# Bleu | Product engineering for work that's hard to delegate Bleu helps software companies and DeFi teams take ownership of stuck engineering work: legacy rebuilds, integrations, new product areas, and production AI features. **Language**: English **Website**: https://bleu.builders/en --- ## Home ### Built to own the problem, not just the ticket. When critical work is stuck and your team can't afford to keep managing it, Bleu steps in to understand the real problem and own it end to end. For software companies and DeFi teams with work that's hard to delegate - Not staff augmentation - You don't need a perfect spec to start - Production-ready delivery, documentation, and handoff - **CoW Protocol**: high-context product and tooling work - **Perk**: ongoing product development partner - **Silo Finance**: data infra owned without pulling core team off protocol > "You don't wait for a super complete specification. You build the plan on your own and do research." > — Sasha, CoW Protocol > "Easy to work with. Take ownership, which is not easy. Can be trusted to deliver, also not easy." > — Siros, Silo Finance > "What you wished your last team was." > — Taylor, Perk ### When teams bring in Bleu - **Something important has been stuck for too long.** — An integration, migration, legacy surface, or product area that keeps surviving the roadmap. - **The core team could do it, but not without slowing something more visible.** — The work matters. Pulling senior people into it means something else stops. - **The work needs judgment, not just execution.** — The problem is still taking shape. Someone has to understand it and keep moving without constant direction. - **You want a clear recommendation before committing to the wrong fix.** — Build, fix, re-scope, or stop. You want a diagnosis, not a proposal. ### What owning the problem looks like The request is often only part of the problem. - **What most vendors do**: Build the SDK they were asked for. Ship the ticket. Mark the work as done. - **What Bleu finds first**: The SDK wasn't the real blocker. The API exposed too much internal complexity, the docs assumed context outside developers didn't have, and the onboarding path didn't exist. - **What Bleu actually ships**: A reworked API surface, the missing docs, and tooling around the onboarding path. Integrations started moving because the whole handoff became usable — not just the SDK. ### Proof in production - **30+** months with CoW Protocol - **16+** initiatives shipped - **3+** years with Perk - **3** active partnerships over 1 year - **CoW Protocol**: SDKs, tooling, integrations, governance and user-facing product work across the ecosystem. - Result: CoW offloaded high-context ecosystem work without slowing the core team or adding management overhead. - Duration: 30+ months (ongoing) - **Perk**: Product, design, engineering, infra, and platform evolution across multiple launches. - Result: Perk kept shipping launches while strengthening the platform underneath — without managing another vendor. - Duration: 3+ years (ongoing) - **Silo Finance**: Subgraphs, multichain APIs, data validation, and the product layer built on top of them. - Result: Silo gained production-grade data infrastructure without diverting the core team from protocol work. - Duration: 1+ year (ongoing) ### How working with Bleu works The point is to reduce load, not create another team you have to manage. - Stop pulling senior people into neglected work - Stop managing a vendor through every unknown - Stop guessing whether the rebuild is even the right move - **You don't need a perfect spec to start**: Bleu resolves what it can without turning every unknown into a dependency. - **Updates when they matter, not constant overhead**: You get visibility into progress and decisions, not a daily standup to run. - **The work ends in production-ready shape**: Code, documentation, and handoff. Maintainable after us. - Assessment - Sprint - Squad - Unclear problem → Assessment - Clear bounded problem → Sprint - Ongoing critical area → Squad ### When Bleu fits #### Bleu is a good fit when - A real problem is stuck, fragile, or outside core team bandwidth - You need a team that can work from partial context and fill in the gaps - The work needs judgment, not just hands - You want a clean handoff when the job is done #### Bleu is not the right fit when - You want staff augmentation against a fixed spec - You need to approve every step before the team can move - There is no clear owner on the client side - The cheapest delivery option matters more than the quality of the outcome ### How we engage Different entry points. Same level of ownership. #### Items #### Assessment A technical and product read on what's going on, what is causing drag, and what the cleanest next step is. #### Sprint A bounded engagement for a problem that needs end-to-end ownership: integrations, migrations, legacy rebuilds, production features, internal tooling. #### Squad A stable team taking responsibility for a product area that matters. ### Sprint #0 We select one project per month to start for real. One week of work, on us, before any contract. One project per month --- ## Assessment | Technical clarity in 1-3 weeks | Bleu A focused technical and product assessment for teams that know something is stuck, fragile, or harder than it should be. ### Technical clarity in For teams that know something is off and want a clear read before committing to the wrong fix. ### When the right next step isn't obvious A system feels fragile. Delivery is slower than it should be. A legacy surface keeps causing drag. A proposed fix sounds expensive, but nobody is fully sure it addresses the real problem. You don't need vague advice. You need a clear diagnosis and a recommendation you can act on. ### Bleu Assessment In 1-3 weeks, we: - Understand the business and technical context - Review the relevant code, architecture, and delivery shape - Identify the actual sources of drag or risk - Recommend the cleanest next step - Outline a practical path forward, with or without Bleu ### Best fit if you... - Know a technical problem is affecting delivery or stability - Want an outside read before committing to a rebuild, migration, or new initiative - Need clarity on whether to fix, re-scope, replace, or stop - Prefer a direct recommendation over a long diagnostic exercise ### What typically comes next Assessment is useful when the next move is still unclear. From there, the path is usually obvious: - Assessment → Sprint: the problem is clear enough to take on as a bounded engagement - Assessment → Squad: the work is ongoing and needs a stable team owning it over time - Assessment → internal execution: you keep the work in-house and use the recommendation to move faster ### Need a clear read before deciding? --- ## Sprint | A real technical problem solved in 2-8 weeks | Bleu A bounded engagement for integrations, migrations, fragile systems, legacy rebuilds, or production features that need end-to-end ownership. ### One real problem solved in Best fit when the problem is real, bounded, and overdue — but still needs judgment to get right. ### When a problem keeps surviving the roadmap Everyone knows the issue matters. Nobody has owned it. It sits between teams, outside the roadmap, or behind higher-visibility work. Meanwhile it keeps creating drag: on delivery, on reliability, on adoption, or on the people who have to work around it. ### Bleu Sprint A Sprint is a bounded engagement where Bleu takes ownership of one concrete problem and gets it to a usable outcome. - Ownership from discovery through delivery - Clear scope and decision points - Working result, documentation, and handoff ### Typical Sprint work - **Integrations**: Third-party APIs, protocol integrations, internal system handoffs, data flows - **Legacy surfaces**: Areas nobody wants to touch, but everybody depends on - **Migrations**: Frontend migrations, stack upgrades, architecture transitions - **Reliability work**: Performance, stability, observability, production hardening - **New product work**: A feature or surface the core team cannot absorb without slowing something more critical ### How it works - **Understand the shape of the problem**: We get enough context to know what we're taking on. - **Set the plan**: We define the path, the scope, and the decision points. - **Execute with visibility**: We keep moving, surface the calls that matter, and avoid creating extra load. - **Deliver in good shape**: The result works, is documented, and can be handed off cleanly. ### Have a problem that's been stuck too long? --- ## Sprint #0 — Bleu Every month, we select one project for a week of real work before any contract. Apply. ### Sprint #0 Every month, we select one project to start for real, before signing anything. ### Not for everyone. Once a month, we identify a project with real potential and propose starting before formalizing anything. One week of concrete work, on us. The client leaves with a running project, milestones, and a locked plan. We leave with real context. Both sides know if it makes sense to continue together. ### What we deliver in the week - **Project running**: Repo, environment, and initial config set up. First tasks done. - **Plan locked**: Milestones and budget detailed within your available budget. - **Next steps**: Everything ready to execute, with us or whoever you choose. ### What we look for - Real technical complexity — context takes time to absorb - A committed client with a defined budget - Scope still forming — both sides benefit from starting before everything is locked on paper - Clear fit — we believe in the problem and the team ### What's NOT included - Not a complete project — it's the kickoff - Doesn't include production-ready code - Doesn't replace a formal contract for full execution ### Want to be considered for the next Sprint #0? --- ## Squad | A team owning a product area | Bleu A stable team taking responsibility for a product or infrastructure area over time. ### A team owning a critical part of your product Best fit when the work is ongoing, tightly connected to the system, and too important to keep staffing ad hoc. ### When the work is not really a project Some problems are too large, too connected, or too persistent to treat as a short engagement. They need a team that can stay close to the system, make decisions with context, and keep improving the area over time. ### Bleu Squad A stable team with ownership of a product or infrastructure area your users depend on. - Not staff augmentation - Stable ownership over time - Continuous product and engineering progress without heavy management overhead ### What a Squad typically owns - A product surface or workflow your users depend on - A critical internal system or data layer - The backlog around a neglected but high-value area - The reliability, performance, and maintainability of that area over time - Documentation and clean knowledge transfer as the work evolves ### Need a team to own a product area? --- ## How Bleu works You bring the problem. We understand it, set the path, and take ownership of getting it to a good outcome. ### How working with Bleu works You bring the context. We do the work required to understand the real problem, take ownership, and get it into good shape. ### You bring the context You do not need a perfect spec. We need enough context to understand what matters and where the drag is showing up. - A focused first call - Business and technical context - What is blocked, fragile, or hard to delegate ### We get close to the work We review the relevant systems, talk to the right people, and figure out what the problem depends on. - Access to the relevant code and environments - Conversations with the people closest to the work - Fast context-building, not endless discovery ### We define the path We tell you what we think the real problem is, what we would do about it, and what kind of engagement fits best. - Clear recommendation - Scope and tradeoffs made explicit - No vague proposal language ### We take ownership Once the work starts, our job is to reduce load on your side, not create more of it. - Regular async updates - Direct access to the people doing the work - Escalation only when a decision needs you ### We leave it in good shape The outcome has to work in production and be maintainable after us. - Production-ready delivery - Technical documentation - Clean handoff to your internal team when needed ### Ways to work with Bleu Each engagement has a different entry point and shape. The path depends on where you are. - **Assessment**: A technical review of an existing system. Produces a clear diagnosis: what the current state is, where the risk is, what needs to change, and in what order. - **Sprint**: A focused delivery on a specific, well-defined problem. A stuck integration, a feature under pressure, a component that needs to be rebuilt. - **Squad**: An embedded product team — PM, design, engineering — working inside your product on an ongoing basis. For when the work is continuous and context matters. ### Ready to start? --- ## Team: Who makes Bleu Senior engineers who've solved problems like yours before. Meet the team. ### Who makes Bleu Senior engineers with real experience solving complex problems. ### 0 ### 1 ### 2 ### 3 ### Tech Stack - **Backend** - **Frontend** - **Infra** - **Data** ### Want to meet the team? --- ## About Bleu Bleu works with teams that need ownership and judgment over their technical work, not just task execution. ### We take ownership of work that's hard to delegate Bleu works with software companies and DeFi teams when the job calls for understanding the problem, making good decisions, and carrying it through. ### Why Bleu exists Bleu came out of a pattern we kept seeing: critical technical work getting delayed, over-scoped, or built around the wrong problem. The issue usually was not effort. It was ownership. Somewhere between the request and the build, the real problem stopped being examined closely enough. Bleu was built to close that gap. The job is to understand what needs to happen, do the work well, and leave it strong enough to hold up in production. ### What we believe A lot of expensive software work starts from a request that is only partially right. That is normal. What matters is catching it early. When Bleu takes something on, we do not treat the brief as untouchable. We question the problem, understand what success depends on, and move from there. That usually leads to better decisions, less wasted motion, and work that does not need to be rethought a month later. ### How we maintain trust Trust isn't declared at the start of a relationship. It accumulates through practices that make the work visible and the accountability clear. - **Status without prompting**: The team initiates updates. Briefly, regularly, at the level of detail that matters. Clients should not have to ask. - **Named owner per engagement**: Every engagement has one person accountable for the outcome. The client knows who that is. That person is not rotated out without a clear handoff. - **Leadership visibility**: For longer engagements, a founder or senior lead checks in at the cadence the engagement warrants. Not because something broke, but to confirm the work is still pointed at the right target. - **Honest allocation visibility**: When capacity changes because something took longer or scope evolved, the client hears about it before it affects delivery. Surprises in allocation are avoidable. - **A real QA bar**: Anything that goes in front of users meets the standard a product company would hold its own engineers to. Shipping something that breaks or looks unfinished reflects on the outcome and the team behind it. - **Recentering when something feels off**: When check-ins cover the same ground, when the brief seems to have shifted, when the team is executing but the outcome feels uncertain. Bleu initiates that conversation rather than waiting. ### Who leads Bleu José built Bleu after seeing the same failure mode too many times: the work was technically strong, the team was capable, but somewhere between the request and the build, the real problem stopped being examined closely enough. The fix was not better process. It was building a company where the people doing the work treat the outcome as their responsibility, and have enough context and judgment to catch problems early, before they become expensive. What Bleu does is simple to describe: demanding technical work, done well, with clarity about what fits and what doesn't. ### What working with us feels like - **We don't overcomplicate**: If we can't explain why we're doing something, we probably shouldn't be doing it. - **We care about the outcome**: Not performatively. We say when something feels off, even when it would be easier not to. - **We integrate, we don't hover**: We learn the product well enough to make decisions with context, not from the outside. - **We take responsibility for the outcome**: Tickets matter. But the point is whether the thing works when it reaches real users. ### How we work Direct, thoughtful, and technically serious. We care about quality, tell the truth when something feels off, and try to make life easier for the team on the other side. ### Numbers that speak - **50+** Projects delivered - **6** Years in operation - **$5M+** In delivered value - **100%** Remote since day one ### Security & Compliance We work with sensitive data and critical systems. We take security seriously. - NDAs signed on all projects - Minimum necessary system access - Mandatory code review - Industry-standard security practices - Security testing on critical systems ### Curious if we can help? --- ## Bleu — DeFi Protocol Engineering Product engineering for DeFi protocols. SDKs, frontends, integrations, analytics, and infrastructure, owned end-to-end by a team that's been shipping in DeFi since 2023. ### Your protocol ships faster when someone owns the work. We work with DeFi protocols that have work their core team can't cover. SDKs, frontends, integrations, analytics, data infrastructure. Taken off your plate and owned end-to-end. ### 30+ months in DeFi. All partnerships still active. We don't do one-off projects. Every protocol we've worked with has kept working with us because the work keeps compounding. - **30+** months with CoW Protocol - **16+** CoW initiatives shipped - **3+** years with Perk - **1+** year with Silo Finance - **9** Balancer grants completed - **4** Optimism grants delivered ### Who we work with - **Established protocols with a growing surface area**: You have a core product and a core team. But SDKs need updating, integrations pile up, frontends for new features don't get built, and developer tooling stays on the backlog. You need a team that can own these areas without constant direction. - **Protocols that need something shipped fast**: A Safe app, a hook dApp, a migration, a dashboard. Scoped work that needs DeFi-native engineers who can start without a month of onboarding. - **Teams that want a clear technical read before deciding**: You're evaluating architecture, considering a migration, or deciding whether to build vs. integrate. You want someone who understands protocol-level tradeoffs and will give you a straight answer. ### What we take on - **SDKs & Developer Tooling**: Python SDKs, TypeScript SDKs, framework-agnostic libraries, solver templates, documentation. Built for adoption and correctness. - **Protocol Frontends**: Safe apps, hook dApps, AMM deployers, dashboards, analytics. Production-ready interfaces for protocol functionality. - **Integrations & Infrastructure**: API integrations, subgraph deployments, data pipelines, monitoring, performance testing. The connective tissue protocols need but can't prioritize. ### What we own end-to-end - Smart contract interaction layers and SDK development - Frontend applications (React, Next.js, Safe apps) - Data infrastructure (subgraphs, indexers, Dune dashboards) - Developer experience (documentation, onboarding, tooling) - Testing infrastructure (load testing, performance benchmarking) - Bot development (monitoring, alerts, MEV detection) ### Protocols we've worked with - **CoW Protocol**: - **Silo Finance**: - **Balancer**: - **Optimism**: - **Morpho**: #### Quote > "You don't wait for a super complete specification. You build the plan on your own and do research." ### Have protocol work that needs ownership? --- ## Contact Bleu Bring us the problem. We'll tell you whether Bleu is a fit and what the cleanest next step is. ### Bring us the problem Tell us what is stuck, fragile, or hard to delegate. We'll tell you whether it makes sense for Bleu to help. ### Schedule a call 15 minutes to understand your problem and see if it makes sense to work together. ### Send an email Prefer writing? We respond within 24 business hours. ### What to expect - We respond within 24 business hours - The first call is for context, not a pitch - We will tell you directly if Bleu is not the right fit - If there is a fit, we'll recommend the cleanest next step --- ## CoW Protocol Case Study: Technical Partnership | Bleu Ongoing technical partnership across developer tooling, user interfaces, and governance infrastructure. ### 16+ initiatives shipped over 30 months without adding management overhead. CoW offloaded high-context ecosystem work — SDKs, tooling, governance, user-facing product — so the core team stayed focused on the protocol. ### Context CoW Protocol is a core piece of DeFi infrastructure, providing MEV protection, programmable orders, and AMMs used by a growing ecosystem of developers, integrators, users, and DAO participants. As the protocol matured, the main challenge shifted from correctness to accessibility and adoption. Powerful infrastructure existed, but using it—or building on top of it—required deep protocol knowledge and specific technical choices. - Developers outside the JavaScript ecosystem couldn't easily integrate - JS/TS teams were locked into Ethers.js v5 - Non-technical users couldn't deploy or manage AMMs without touching contracts - Governance data was fragmented across multiple tools ### Our role We partnered with CoW on initiatives that were high leverage but hard to delegate—work requiring deep protocol understanding, strong frontend and tooling expertise, and the ability to operate with minimal specification. #### Quote > "You're proactive. You don't wait for a super complete specification… with you it was easy." ### What changed Over time, CoW's ecosystem became easier to work with—without forcing users or developers into a single stack, workflow, or level of technical expertise. - Developers can now integrate using the tools and languages they already use - Frontend teams gained flexibility and reduced technical debt - Users can deploy and manage AMMs through familiar interfaces - The DAO has a clearer, more transparent view of governance participation ### Areas we took ownership of #### Opening the protocol to new developers CoW's JavaScript SDK left Python developers without a native way to integrate. - **Python SDK**: CoW's existing SDK ecosystem was JavaScript-first, leaving Python developers without a native integration path. • Built a Python SDK from scratch with full feature parity with the JS SDK • Enabled order management, on-chain queries, and smart contract integration • Opened the protocol to a developer ecosystem where Python dominates - **Framework-agnostic TypeScript SDK**: The existing TS SDK was tightly coupled to Ethers.js v5, forcing all developers into a single stack. • Re-architected the SDK around Viem • Created adapters for Ethers v5, Ethers v6, Viem, and Wagmi • Reduced bundle size by 32% (1447kb → 981kb) • Shipped in 3 months, unlocking long-term flexibility #### Reducing frontend friction and technical debt CoW Swap's frontend relied on Web3React - an aging, largely unmaintained library causing wallet issues. - **Frontend migration**: CoW Swap's frontend relied on Web3React - an aging, largely unmaintained library that caused wallet connection issues and slowed development. • Leading a full migration to Wagmi + Viem • Fixing wallet connection flakiness and enabling native EIP-6963 (multi-wallet) support • Reducing maintenance burden and onboarding friction for new developers • Expected completion: Q1 2025 - **CoW AMM Deployer (Safe App)**: CoW launched AMM contracts with strong community demand - but no interface to use them. • Delivered a production-ready Safe App in just 2 weeks • Enabled users to create, manage, and stop AMMs without writing code • Became the primary interface at a critical adoption moment, while the core team stayed focused on protocol work #### Making protocol complexity legible to users and the DAO Abstract concepts made concrete. Governance made transparent. - **Have I Been MEV'd**: To drive awareness around MEV exploitation and adoption of MEV Blocker, CoW needed a way to make a complex topic immediately understandable. • Built bots for Twitter, Discord, and Telegram to check MEV exposure instantly • Created a web dashboard for address scanning and wallet connection • Turned a complex concept into shareable, viral moments across social platforms - **Governance dashboard**: Governance data lived across Snapshot, on-chain contracts, and fragmented tools. • Built a Dune dashboard consolidating proposals, voting analytics, and delegation data • Enabled the community to track participation, quorum status, and voting power distribution • Gave the DAO a clear, accessible view of governance health ### Why this partnership works CoW's core team focuses on shipping protocol and user-facing features. That leaves a long tail of critical technical work—SDKs, migrations, refactors—that is foundational but difficult to prioritize internally. This partnership works because Bleu operates effectively in that space: taking ownership of ambiguous, high-impact technical work with minimal guidance, while maintaining high code quality and low coordination overhead. #### Quote > "The team worked super fast, operated on minimal guidance and took instruction super well." ### Outcome By offloading hard-to-delegate technical work without adding management overhead, CoW was able to expand its ecosystem, improve developer and user experience, and strengthen long-term protocol maintainability, while keeping the core team focused on advancing the protocol itself. If you're dealing with work that's hard to delegate but hard not to do, this is the kind of partnership we build. > "See how we work" > "Bring us the problem" ### What we've shipped - **cow-py**: Community-built Python SDK and baseline solver for CoW Protocol - **Framework Agnostic SDK**: Restructured SDK architecture with framework-agnostic base packages - **CoW Hooks dApps**: Four production hook applications for post-trade execution - **CoW Swap Frontend Migration**: Migration to Viem + Wagmi, removing legacy dependencies - **Morpho Hooks**: Morpho position management hooks for the CoW Hook Store - **Safe Stop-Loss Orders**: Safe App for multisig stop-loss order management - **Safe Milkman Orders**: Safe App for creating Milkman orders with deferred pricing - **CoW AMM Deployer**: Safe App to create, edit, and manage CoW AMMs - **CoW AMM Expansion**: Oracles, standalone contracts, and expanded AMM functionality - **MEVaporized Bot**: Twitter, Discord, and Telegram bots + dashboard for MEV detection - **Dune Governance Dashboard**: Analytics for proposals, voting, delegations, and delegate analysis - **Programmatic Orders API**: ComposableCoW tracking, flash loan orders, and orderbook integration - **Playground Performance Testing**: Load generation, benchmarking, and metrics visualization suite - **Playground Offline Mode**: Offline development mode with DEX liquidity and test tokens - **CoW Subgraph Redeploy**: Redeployment of CoW Protocol subgraph infrastructure - **Solver Infrastructure**: Python baseline solver, templates, and onboarding tooling --- ## Silo Finance Case Study - Data Infrastructure Partner | Bleu Data infrastructure partner for multichain lending protocol. Building the indexing and data layer everything else depends on. ### Production-grade data infrastructure without pulling the core team off protocol work. Silo needed subgraphs, multichain APIs, and data validation — but couldn't afford to divert the team building the protocol. Bleu owned it end to end. ### Context Silo Finance is a permissionless lending protocol designed for long-tail assets, where each market is isolated to contain risk. As they scaled to V2 and expanded across multiple chains, reliable data infrastructure became critical—not just for internal operations, but for user-facing features like points programs and health factor visibility. The challenge wasn't building the protocol itself. It was building the data layer that everything else depended on: accurate position tracking, real-time computed fields, and multichain consistency. These were foundational but fell outside the core team's bandwidth and expertise. - V2 launch needed an official subgraph without internal bandwidth - Existing points program pipeline had undetected data quality issues - Multichain expansion required unified data access across 4+ networks - Infrastructure costs were growing with each new integration ### Our role We became Silo's data infrastructure partner—owning the systems that sit between on-chain state and the interfaces users and integrators rely on. This meant building from scratch, auditing existing pipelines, and expanding coverage across chains—all with minimal specification, operating as an embedded technical team rather than a vendor. #### Quote > "Easy to work with. Take ownership, which is not easy. Can be trusted to deliver, also not easy. If we go through all the partners we've worked with, you guys are probably the best." ### What changed Silo now has production-grade data infrastructure across their entire stack—and reduced infrastructure costs by 80% in the process. - V2 subgraph provides reliable, auditable on-chain data for the protocol's core markets - Points program launched without corrupted distributions—errors caught before reaching users - Integrators and users can access real-time health factors and USD pricing across all supported networks - Infrastructure costs reduced by 80% ### Areas we took ownership of #### Silo v2 Subgraph Silo's v2 launch needed an official subgraph, but building reliable indexing infrastructure wasn't where the core team was focused. - **Official Subgraph**: Silo's v2 launch needed reliable indexing, but building production-grade subgraphs wasn't a core team priority. • Designed and deployed the mainnet subgraph for Silo v2 • Provided the foundation for all downstream data access—dashboards, integrations, internal tooling • Auditable on-chain data for lending markets and risk management #### Data Audit & Validation Before the points program launched, we audited the existing pipeline to catch hidden data quality issues. - **Points Program Validation**: Silo's existing points program pipeline had issues that weren't immediately visible. Before launch, we validated the entire pipeline. • Built an independent source of truth using Ponder, validated against on-chain data with 100% accuracy • Identified rounding errors, missing Transfer events, and ~36% duplicate records in PositionSnapshot queries • Delivered detailed root cause analysis and impact assessment—preventing corrupted distributions • Ensured user rewards remained accurate and trust-preserving #### Multichain API & Infrastructure As Silo expanded across Arbitrum, Optimism, Base, and Sonic, they needed a unified data layer. - **Multichain API**: As Silo expanded across Arbitrum, Optimism, Base, and Sonic, they needed a unified data layer that could compute real-time fields on demand—not just mirror on-chain state. • Health factor and bad debt calculations on-demand • USD pricing with integrations for Chainlink, Pyth, and RedStone oracles • Unified interface across all supported networks • 80% infrastructure cost reduction through optimization and consolidation ### Why this partnership works Silo's core team focuses on protocol development and expansion. That leaves critical infrastructure work—subgraphs, data validation, multichain APIs—that is foundational but difficult to prioritize internally. This partnership works because Bleu operates effectively in that space: taking ownership of high-impact technical work with minimal guidance, while maintaining production-grade quality and low coordination overhead. #### Quote > "By offloading data infrastructure work, we were able to launch with confidence and focus on what we do best - building the protocol." ### Outcome By offloading data infrastructure without adding coordination overhead, Silo launched v2 with reliable indexing, caught critical errors before they impacted users, and expanded multichain coverage—all while keeping the core team focused on the protocol itself. Plus, they cut infrastructure costs by 80% in the process. If you're dealing with infrastructure work that's hard to delegate but hard not to do, this is the kind of partnership we build. > "See how we work" > "Bring us the problem" --- ## Perk: Product Partnership Case Study | Bleu Ongoing product development partnership across loyalty infrastructure and customer engagement platforms. ### 3+ years shipping launches while strengthening the platform underneath. Perk needed to keep shipping product without letting tech debt and platform gaps slow them down. Bleu became the product development partner that could do both. ### Context Perk is a loyalty platform with an ambitious vision: transform loyalty programs from static point systems into dynamic, multi-channel engagement engines. The goal was to automatically turn any brand interaction — purchases, social shares, referrals, content engagement — into rewardable moments that drive measurable behavior change. The challenge wasn't just technical. Perk was entering a crowded market where most loyalty solutions felt bolted-on rather than built-in. What was at stake: proving that loyalty could be both flexible and powerful, that brands could launch real programs without enterprise-level budgets or timelines. - Competitive market with solutions that feel bolted-on rather than built-in - Need to validate ideas quickly with real users - Complex reward logic and receipt processing at scale - Balance flexibility with power and simplicity ### Our role We became Perk's core product development partner — moving quickly enough to validate ideas with real users, deep enough in the system to handle reward logic and receipt processing at scale, and involved enough in product decisions to shape direction rather than just execute tickets. #### Quote > "What you wished your last team was." ### What changed Perk's platform now operates at a different scale without engineering becoming a bottleneck. Work that used to require heavy custom effort now moves faster because the platform can absorb new rules, channels, and interfaces without being rethought each time. That gives Perk room to validate ideas with real users before committing to larger roadmap bets. - 1.6M+ tracked actions in 2025 - 780k+ completed challenges - 220k+ rewards processed - 100k+ receipts verified - Time-to-launch reduced from months to days ### 0 ### 1 ### 2 ### 3 ### Areas we took ownership of #### Core Loyalty Engine Built the foundational infrastructure that powers Perk's flexible reward mechanics. - **Challenge Completion Logic**: Designed challenge completion logic that processed 780k+ completed challenges, handling edge cases around timing, eligibility, and reward distribution. - **Rules Engine**: Architected a rules engine flexible enough to support any brand's loyalty model while maintaining performance as action volume scaled to 1.6M+ tracked actions. - **Real-time Reward Claiming**: Enabled real-time reward claiming with 220k+ rewards processed, balancing instant gratification with fraud prevention. - **Cloud Infrastructure**: Design, implementation and load testing of cloud infrastructure supporting 10k+ concurrent users. #### Receipt Processing & Verification Perk's differentiator was bringing offline purchases into the loyalty loop. - **OCR and Validation Pipeline**: Developed OCR integration and validation pipeline that processed 100k+ receipts with automated item extraction and eligibility verification. - **Fraud Detection**: Implemented fraud detection patterns while keeping the user experience frictionless — customers get validated quickly, brands get protected automatically. #### Multi-Channel Action Tracking Loyalty only works if it's everywhere your customers are. - **Platform Integrations**: Integrated with e-commerce platforms, social media APIs, and referral systems to capture actions automatically. - **Tracking Architecture**: Designed tracking architecture that onboarded 100k+ new active users while maintaining data consistency across channels. - **Attribution Logic**: Built attribution logic ensuring users get credit for their actions regardless of where they happen. #### Admin & End-User Interfaces Built the interfaces that give Perk's clients visibility into what drives customer behavior. - **Real-time Dashboards**: Real-time dashboards showing challenge completion rates, reward redemption patterns, and engagement trends. - **White-label Program Builder**: White-label program builder enabling Perk to serve clients of different sizes and industries without rebuilding core infrastructure. #### Developer Platform & Extensibility Perk needed to be more than a closed system — it had to fit into clients' existing tech stacks. - **Comprehensive REST API**: Comprehensive REST API letting brands handle rewards, challenges, user data, and sync loyalty state with their own systems — enabling use cases from custom mobile apps to point-of-sale integrations. - **Real-time Webhook Streaming**: Real-time webhook streaming for challenge completions, reward claims, and user milestones. - **Aptos Integration**: Aptos integration enabling on-chain reward distribution and NFT-based loyalty mechanics. ### Why this partnership works Perk's team focuses on client relationships, go-to-market, and product direction. Deep product development work — infrastructure scaling, new integrations, self-service tooling — is essential for growth, but difficult to prioritize alongside serving existing clients. This partnership works because Bleu operates as an embedded product team: taking ownership of both strategic and execution work, thinking through downstream implications, and shipping production systems that don't require constant oversight. #### Quote > "You guys do such a nice job of thinking through what that means, where that impacts the rest of the platform, user experience, what the client's looking for. I'm not up at night worried that it's going to break." ### Outcome By taking ownership of product development without adding management overhead, Bleu helped Perk reduce launch time, build self-service tools that remove bottlenecks, and scale infrastructure to handle real volume. The core team stays focused on the business while the platform keeps pace with what the market is asking of it. ### Detailed Case Studies - **Legacy Takeover**: Taking over an outsourced codebase with no CI, no tests, no PR reviews — and modernizing it without downtime. - **AI-Powered Onboarding**: Production LLM pipeline that generates fully configured loyalty programs from a brand's website URL. - **M&M's Fun Club**: Enterprise-grade infrastructure rebuild and global brand launch with 99.984% uptime. - **Pedigree One True Loyalty**: Platform evolution from loyalty engine to full CRM with marketing automation and internationalization. If you're dealing with product work that's hard to delegate but hard not to do, this is the kind of partnership we build. > "See how we work" > "Bring us the problem" --- ## M&M's Fun Club powered by Perk | Bleu Scaling a loyalty platform from startup to enterprise-grade for a global brand launch. ### From 500ms endpoints to enterprise-grade in time for a global brand launch. M&M's needed a loyalty platform ready for thousands of concurrent users. The existing infra wasn't close. Bleu rebuilt the performance layer, monitoring, and incident response before the hard launch date. ### Context In early 2025, Perk landed their biggest opportunity yet: M&M's wanted to launch "M&M's Fun Club" – a customer engagement platform that would go live in February 2025. The problem was the existing infrastructure wasn't ready for primetime. Endpoints were responding in 500ms+. The codebase was riddled with n+1 queries and memory allocation issues. There was no comprehensive monitoring, no incident response system, and no confidence the platform could handle the traffic a major global brand would bring. With thousands of concurrent users expected and a hard launch date approaching, Perk needed more than incremental improvements – they needed a complete technical transformation. - Endpoints responding in 500ms+ with n+1 queries and memory allocation issues - No comprehensive monitoring or incident response system - Infrastructure not ready for global brand traffic - Hard launch date in February 2025 ### Our role We took ownership of the technical foundation. Rebuilding infrastructure from the ground up while shipping the features M&M's needed for launch. The stakes were clear: if the platform failed during launch, it wouldn't just hurt Perk's relationship with M&M's — it would undercut confidence that Perk could operate at enterprise expectations. #### Quote > "These aren't small clients. We have real repercussions as a company if things go wrong, even financial penalties for downtime. It's nice that I'm not up at night worried it's going to break." ### What changed The platform moved from fragile under load to stable in production. Response times dropped from 500ms+ to under 50ms. Load testing validated 10,000 concurrent users, and since launch the platform has maintained 99.984% uptime. - M&M's launched on schedule in February 2025 with zero downtime during the critical launch period - Response times from 500ms+ to under 50ms - 10,000 concurrent users validated through load testing - 99.984% uptime since launch ### Areas we took ownership of #### Infrastructure Rebuild Replaced manual deployments with Infrastructure as Code using AWS CDK. - **Infrastructure as Code**: Replaced manual deployments with Infrastructure as Code using AWS CDK. Implemented enterprise security standards from day one and designed for horizontal scalability to handle unpredictable traffic spikes. #### Performance Optimization Achieved 90% latency reduction through systematic optimization. - **Systematic Optimization**: Achieved 90% latency reduction through systematic optimization – eliminating n+1 queries, adding strategic indexes, and resolving memory allocation issues throughout the codebase. Validated everything through rigorous load testing until the application handled 10,000 concurrent users smoothly. #### Receipt Scanning System Built the core engagement mechanic for the platform. - **Receipt Scanning**: Built the core engagement mechanic: customers submit receipts containing M&M's products, the system processes and validates them automatically, and users redeem rewards for M&M's branded merchandise. #### Gaming Challenges Implemented interactive mini-games integrated into the platform. - **Gaming Challenges**: Implemented interactive mini-games integrated into the platform with a point-based reward system and exclusive M&M's-themed challenges. #### Notification System Platform-wide notifications and targeted user messaging. - **Notifications**: Added a notification system for platform-wide messaging and targeted user communications, giving M&M's another way to react to user activity during launch. #### Custom Frontend Built a dedicated frontend for the M&M's experience. - **M&M's Frontend**: Built a dedicated frontend for M&M's on top of the rebuilt platform, giving the brand its own customer-facing experience without forking the core system. #### GraphQL API Designed a new API specifically for the M&M's frontend. - **New GraphQL API**: Designed a new GraphQL API specifically for M&M's frontend, with optimized queries that eliminated the n+1 problems plaguing the legacy REST endpoints. #### Data & Analytics CSV processing engine and admin dashboards. - **Data Processing**: Built a CSV processing engine that exports millions of rows in seconds, plus admin dashboards for metrics including access patterns, retention data, and stock information. #### APM & Monitoring Established health check routines and proactive issue detection. - **Health Check**: Established weekly health check routines monitoring infrastructure, email delivery, and response times – enabling proactive issue detection before users notice problems. #### Security Hardening Implemented OTP system with protection against attacks. - **OTP System**: Implemented a one-time password system with protection against brute-force attacks, rainbow table attacks, and birthday problem vulnerabilities. #### On-Call & Incident Response 24/7 coverage with senior engineers on rotation. - **On-Call Rotation**: Set up 24/7 coverage with senior engineers on rotation, ensuring 100% incident response coverage with rapid resolution protocols. #### Webhook System Configurable webhooks for real-time event emission. - **Webhooks**: Built configurable webhooks for real-time event emission, enabling synchronization between M&M's infrastructure and Perk. ### Why this partnership works M&M's launch required both speed and precision — there was no room for the platform to give way under pressure, and no time to build slowly. Bleu operated as an extension of Perk's team, taking full ownership of technical decisions while Perk focused on the client relationship and launch coordination. ### Outcome The M&M's Fun Club launch established that Perk could support a global-brand rollout under real operating constraints. By rebuilding infrastructure, improving performance, and putting reliability practices in place before launch, the team shipped in February 2025 without downtime and left Perk with a foundation it could reuse on later enterprise work. If you're dealing with infrastructure challenges or need to scale for a major launch, this is the kind of work we do. > "See the full Perk case study" > "Bring us the problem" --- ## Pedigree One True Loyalty powered by Perk | Bleu Transforming a loyalty platform into a full-featured CRM with marketing automation and international reach. ### From loyalty program to full CRM with marketing automation across multiple countries. Pedigree needed more than a loyalty platform — they needed audience segmentation, email campaigns, internationalization, and a custom frontend. Bleu expanded what the platform could do. ### Context After M&M's Fun Club launched successfully, Pedigree came to Perk with a bigger vision. They didn't want just a loyalty program – they wanted a complete customer relationship management platform that could launch across multiple countries and give their marketing team real power. Pedigree needed everything we'd built for M&M's, plus sophisticated marketing automation, email campaigns, audience segmentation, and internationalization. And they wanted a completely custom frontend that matched their brand perfectly. This wasn't about adding features to an existing product. It was about expanding what a loyalty platform could be. - Complete CRM, not just loyalty program - Sophisticated marketing automation and email campaigns - Advanced audience segmentation - International multi-country support - Completely custom frontend ### Our role We extended Perk's core platform into territory it hadn't touched before — marketing automation, campaign management, and multi-country support. At the same time, we built Pedigree a completely custom frontend from scratch. The goal was to give Pedigree the marketing capabilities they needed without introducing a separate operational system or rebuilding the platform for each market. #### Quote > "There's a level of professionalism and consistency in that output being high quality and trustworthy. Instead of just saying yes and doing the thing, you make sure it's lining back to its purpose. Being mindful of how all that impacts each other. That's really unique." ### What changed Perk's platform moved beyond loyalty into lifecycle tooling brands can operate themselves. Teams can now automate engagement, segment audiences, run campaigns, and localize the experience across multiple countries from the same core platform. - Platform evolved from loyalty engine to full CRM - Brands can automate entire customer engagement strategy - Targeted campaigns at scale with localized experiences - Capabilities built into core – available to all future partners ### Areas we took ownership of #### Event-Driven Automation Engine Extended Perk's core to respond automatically when users take actions. - **Automation Engine**: Extended Perk's core to respond automatically when users take actions — completing a challenge, submitting a receipt, hitting their birthday. Admins can create time-based automations (birthday rewards, 'we miss you' messages after inactivity) and action-based ones (2x points for 24 hours after signup). Workflows can send emails, award points, apply multipliers, and more, enabling each brand to create exactly the engagement experience they want. #### Visual Email Template Builder Built a drag-and-drop email builder from scratch. - **Email Builder**: Built an email builder from scratch so admins could create templates without touching code, using double-brace syntax like {{user_name}} to personalize messages. #### Campaign Management & Audience Segmentation The platform now lets admins slice their user base however they want. - **Campaigns & Segmentation**: Admins can define audience segments — most active users, recent signups, or people who have not engaged in weeks — and run the same workflows used by automations. That turned campaign execution into part of the platform rather than a separate manual process. #### Internationalization Framework Comprehensive i18n support for multi-country launch. - **Internationalization**: Since Pedigree planned to launch across multiple countries, we built comprehensive i18n support. The platform handles multiple languages, regional customization, different timezones and currencies – everything needed to go global without rebuilding for each market. #### Custom Next.js Frontend Completely custom frontend from scratch capturing Pedigree's brand identity. - **Custom Frontend**: Pedigree wanted a distinct customer-facing product. We built a completely custom frontend from the ground up — fast, responsive, aligned to the brand, and connected to Perk's backend infrastructure. ### Why this partnership works Pedigree's requirements pushed Perk's platform into new territory. Rather than building one-off features for a single client, we designed systems that would benefit the entire platform. Bleu took ownership of both the strategic architecture decisions and the implementation details, turning a client request into a platform evolution. ### Outcome What started as a loyalty request became a platform transformation. Pedigree got the CRM capabilities needed to run marketing across multiple countries. Perk got an automation engine, campaign system, and internationalization framework it can keep using with future partners because the work was built into the core platform. If you need to transform an existing platform or build CRM capabilities, this is the kind of work we do. > "See the M&M's Fun Club case study" > "See the full Perk case study" > "Bring us the problem" --- ## Perk: Legacy Takeover & Stabilization | Bleu Taking over an outsourced codebase with no CI and no tests, and modernizing it incrementally while shipping product. ### From outsourced codebase to production-grade in 90 days — without stopping delivery. Perk inherited an outsourced Rails monolith on GitLab. No CI. No PR reviews. No tests running. Bleu took ownership, rebuilt the engineering process, and shipped product simultaneously. ### Context Perk's codebase had been built by an outsourced team in India and handed off. The team was gone. There was no continuous integration, no pull request reviews, no test suite running. People pushed directly to branches and merged without review. Deployments were manual. The question wasn't whether to fix this — it was whether a new team could take ownership of someone else's production system, understand it deeply, and modernize it without breaking what was already working. - No CI pipeline — code shipped without automated checks - No PR reviews — direct branch merges without oversight - No test suite running — no safety net for changes - Manual deployments — each release a manual, error-prone process - Previous team gone — no knowledge transfer, no documentation ### Our role Instead of just fixing bugs on top of the existing setup, we rebuilt the engineering process while shipping product. Both happened at the same time — stabilization was not separate from delivery. #### Quote > "What you wished your last team was." ### What changed Within 90 days, the codebase went from ungoverned to professionally managed. Every change went through CI, got reviewed, and was deployed automatically. And product kept shipping throughout. - Migrated from GitLab to GitHub with branch protection and mandatory reviews - CI pipeline running on every push — tests, lints, builds - Containerized with Docker for consistent environments - Introduced background job processing (Sidekiq) - Integrated TailwindCSS for frontend consistency - Linear for project tracking (864 tickets to date) - Shipped social feed challenges, email delivery fixes, and admin auth fixes simultaneously ### Areas we took ownership of #### Engineering Process Rebuilt the development workflow from the ground up. - **Version Control & Code Review**: Migrated from GitLab to GitHub. Set up branch protection rules, mandatory PR reviews, and a merge workflow that ensures every change is reviewed before it ships. - **Continuous Integration**: Built a CI pipeline from scratch. Tests, lints, and builds run on every push. No code reaches production without passing automated checks. - **Containerization**: Containerized the application with Docker, eliminating environment inconsistencies and establishing repeatable builds from development to production. #### Infrastructure Modernization Replaced manual processes with automated, repeatable systems. - **Automated Deployments**: Replaced manual deployments with automated pipelines. Each merge to main triggers a build, test, and deploy cycle with zero human intervention required. - **Background Processing**: Introduced Sidekiq for background job processing, moving email delivery, media processing, and data sync out of the request cycle. #### Product Delivery Shipped features while modernizing — not after. - **Social Feed Challenges**: Built social feed challenge mechanics while the engineering overhaul was underway. Users could post, share, and earn rewards through social interactions. - **Email & Media Fixes**: Fixed email delivery reliability and S3 media link issues that had been degrading the user experience. Stabilized auth flows across the platform. ### Why this matters Most teams won't touch someone else's legacy system. The ones that do usually propose a ground-up rewrite — months of work before any value is delivered. We modernized incrementally: shipped product on day one while rebuilding the foundation underneath. Not a single day of downtime during the transition. Three years later, that codebase is a 16-app monorepo serving Fortune 500 clients with 99.984% uptime. ### Outcome The stabilization set the foundation for everything that followed. The CI pipeline, PR reviews, and automated deployments we put in place in the first 90 days are still running three years later — now processing 12,000+ commits across 1,600+ pull requests. What started as a rescue became a partnership. If you have a codebase you're afraid to touch or a team transition to navigate, this is the kind of work we do. > "See the full Perk case study" > "Bring us the problem" --- ## Perk: AI-Powered Program Onboarding | Bleu Production LLM pipeline that generates fully configured loyalty programs from a brand's website URL. ### From days of manual setup to minutes — powered by production-grade AI. Launching a new loyalty program took days of configuration. We built an AI system that generates a fully configured program from a brand's website URL, with retryable LLM calls, validation, and audit trails. ### Context Every time Perk onboarded a new client, someone had to manually configure the program: upload brand assets, pick colors, create challenges, configure products, set up tiers. This took days per client and required developer involvement for most steps. The goal wasn't to add a chatbot. It was to build AI infrastructure that could reliably analyze a brand's online presence and generate a production-ready loyalty program configuration — automatically. - Program setup took days of manual configuration per client - Required developer involvement for most steps - Brand assets, colors, challenges, products, tiers — all manual - AI output needed to be production-ready, not suggestions - Must work reliably across diverse brand websites ### Our role We designed and built 'Assist' — an AI module that takes a brand's website URL and generates a fully configured loyalty program. Not a prototype. A production system with retryable calls, validation, audit logging, and feature-flagged rollout. #### Quote > "There's a level of professionalism and consistency in that output being high quality and trustworthy. Instead of just saying yes and doing the thing, you make sure it's lining back to its purpose." ### What changed Program setup went from days to minutes. The AI pipeline handles brand analysis, color extraction, challenge generation, product identification, and reward suggestions — all automatically. - Program setup reduced from days to minutes - 9-step onboarding wizard with save/resume capability - Automatic brand analysis from any website URL - Challenge, reward, and tier configurations generated automatically - Feature-flagged rollout for controlled adoption ### Areas we took ownership of #### Website Analysis Pipeline Headless browser captures and analyzes brand websites automatically. - **Headless Browser Capture**: Ferrum (headless Chromium) takes screenshots and captures site content asynchronously. The system navigates the brand's website, identifying pages, products, articles, and social accounts. - **Brand Extraction**: LLM analyzes captured data to extract brand colors with contrast checking, detect brand tone from social media posts, identify products from the store, and pull content for challenges. #### LLM Infrastructure Production-grade AI pipeline — not a wrapper around an API call. - **Retryable LLM Calls**: Custom validation functions verify each LLM output. Failed calls retry with previous-attempt error feedback piped into the retry prompt, improving success rates without manual intervention. - **Audit Trail**: Every LLM call is logged with raw response, validation result, error details, and retry history. Full traceability from input to output. - **Background Processing**: All extraction pipelines run as background jobs via Solid Queue. The UI shows progress in real-time while heavy processing happens asynchronously. #### Program Generation AI output becomes a real, configurable program — not just suggestions. - **Onboarding Wizard**: 9-step wizard with save/resume via onboarding drafts. Each step can be run independently or as part of the flow. Admins review and adjust AI suggestions before finalizing. - **Configuration Output**: Generates branded color palette, challenge configurations (article visits, product quizzes, social feeds), eligible products for receipt scanning, reward suggestions, and tier structure. ### Why this matters This is the difference between 'we integrated an LLM API' and 'we built AI infrastructure that operates reliably under real conditions.' Retryable calls with custom validation. Error feedback piped into retry prompts. Full audit trail. Feature-flagged rollout. Background job processing. Production AI means the system works when nobody is watching, not just during demos. ### Outcome Program setup went from a multi-day manual process to minutes of automated generation plus human review. The AI infrastructure we built — retryable calls, validation, audit logging — is reusable across any future AI features Perk adds to the platform. If you're building AI features into an existing product and need them to work reliably in production, this is the kind of work we do. > "See the full Perk case study" > "Bring us the problem" --- ## Careers Join the Bleu team. Browse open positions and apply directly. ### Work with us Senior engineers who want to own real problems. No ticket factories. No fake deadlines. Hard work on things that matter. ### Application submitted! ### Don't see a role that fits? We're always looking for talented people. Send us your details and we'll keep you in mind for future opportunities. ### Know someone who'd be a great fit? We pay US$ 2,000 for successful referrals (after 90 days). --- ## Cases ### 0 - Frontend - SDKs - Developer Experience - Data Analysis ### 1 - Data Infrastructure - Backend - Multichain ### 2 - Product Development - Infrastructure - Loyalty - CRM ### 3 - Infrastructure - Performance - Enterprise - Receipt Scanning ### 4 - CRM - Marketing Automation - i18n - Enterprise ### 5 - SDKs - Analytics - Tooling ### 6 - Governance - AI - Social ### 7 - DeFi - Smart Contracts - Landing Pages - **CoW Hooks for Morpho**: Unified hook dApp — supply, borrow, repay, withdraw — reviewed by CoW team and integrated into the CoW Swap Hook Store. - **Morpho Vault Deposits**: Co-branded landing page for depositing any token into Morpho Vaults in one gasless transaction via CoW Swap. ### 0 > "See how we work" ### 1 > "Bring us the problem" --- ## Morpho × Bleu: CoW Hooks and Vault Deposit Interfaces Grant-funded CoW Protocol hook integrations and co-branded deposit landing page for Morpho markets. ### CoW Hooks and vault deposit interfaces for Morpho Morpho markets needed composability with CoW Swap. The initial plan was four separate hook dApps, but we realized the operations were interdependent — a fragmented UX. We proposed a single unified interface, got approval, and shipped all four operations plus a standalone deposit landing page. ### The context Morpho needed users to enter and exit lending markets through CoW Swap. The original scope was four separate hook dApps — one per operation. During development, we realized this would create a confusing experience. Many Morpho functions depend on each other, and splitting them meant users would bounce between interfaces for related operations. - Many Morpho functions are interdependent — separated hooks would create a fragmented user experience - Hook dApps needed to pass CoW team review and integrate into the official Hook Store - Users needed real-time LTV calculations and health monitoring during the flow - Transaction safety required Tenderly simulation before execution ### What we delivered Instead of four fragmented hooks, one unified dApp handles everything. The CoW team reviewed and approved all operations, and the hooks are live on CoW Swap's Hook Store. - Unified hook dApp covering supply, borrow, repay, and withdraw in a single interface - All four hooks reviewed and approved by the CoW team - Full integration with CoW Swap's Hook Store — accessible to all CoW Swap users - Real-time LTV calculations and health factor monitoring - Tenderly simulation for transaction safety before execution - Co-branded landing page for depositing any token into Morpho Vaults in one gasless transaction ### CoW Hooks for Morpho Unified hook dApp — supply, borrow, repay, withdraw — reviewed by CoW team and integrated into the CoW Swap Hook Store. ### Morpho Vault Deposits Co-branded landing page for depositing any token into Morpho Vaults in one gasless transaction via CoW Swap. If you need composable DeFi integrations that work end-to-end, this is the kind of work we do. > "See how we work" > "Bring us the problem" --- ## Balancer × Bleu: 9 Grants Across Analytics, SDKs, and Tooling 9 grants delivered for Balancer spanning pool simulation, APR tooling, alerts infrastructure, SDK contributions, and internal balance management. ### 9 grants delivered for Balancer Balancer needed specific tooling that didn't exist — pool simulation, APR calculation, alerting, and SDK gaps. We scoped each piece as an independent grant, built it, shipped it, and moved on to the next. ### What we delivered Each grant addressed a concrete gap in the ecosystem. The tooling is in production, used by the protocol team and integrators. - Pool simulator for modeling AMM behavior and impermanent loss scenarios - APR tooling for accurate yield calculations across pool types - Alerts infrastructure for monitoring pool health and anomalies - SDK contributions improving developer experience for protocol integrators - Internal balance management tools for protocol operations - All 9 grants delivered on scope and on time ### Pool Simulator Interactive tool for modeling AMM behavior, simulating swaps, and analyzing impermanent loss across different pool configurations. ### APR Tooling Accurate yield calculation engine covering weighted, stable, and boosted pool types with real-time data. ### Alerts Infrastructure Monitoring system for pool health, anomaly detection, and operational alerts. ### SDK Contributions Developer experience improvements to the Balancer SDK, making protocol integration smoother for third-party builders. If you have protocol tooling gaps that keep getting deprioritized, this is the kind of work we do. > "See how we work" > "Bring us the problem" --- ## Optimism × Bleu: Governance, Social Graphs, and AI Tooling 4 grants delivered for the Optimism ecosystem spanning governance gamification, Farcaster social graph analysis, AI tools, and tech partner contributions. ### 4 grants delivered for the Optimism ecosystem Optimism needed governance gamification, sybil detection for Farcaster, an AI assistant, and a tech partner program to help OP-native projects ship. We delivered all four — GovQuests with leaderboards and streaks, a social graph sybil model with 0.7+ AUC, a chatbot trained on Optimism governance, and end-to-end development for 8+ projects including Pike, Namespace, and Candide. ### What we delivered Each grant addressed a different layer of the ecosystem. GovQuests gamified governance participation. The sybil model is deployed as a public API and Farcaster Mini App. The AI assistant is live. The tech partner program shipped apps for 8+ teams. - GovQuests: governance gamification platform with leaderboards, streaks, tier system, push notifications, and OP token rewards — all milestones approved - Farcaster Social Graph: sybil detection model achieving 0.7+ AUC, deployed as a public API with a Farcaster Mini App for user reporting - Bleu Assistant AI: chatbot trained on Optimism governance data with Q&A and automated forum discussion summaries - Tech Partner Program: end-to-end design, backend, frontend, and analytics for 8+ OP-native projects — Pike, Namespace, Candide, Silo, and others - All milestones approved and disbursed across all four grants ### GovQuests Governance gamification platform with quests, leaderboards, streaks, tiers, and OP token rewards. Designed to increase delegate participation and make governance accessible to newcomers. ### Farcaster Sybil Detection Social graph-based sybil detection for Farcaster using machine learning and graph algorithms. Public API, Farcaster Mini App, and monitoring dashboard — all open source. ### Optimism AI Assistant Chatbot trained on Optimism governance data. Answers community questions and generates automated summaries of forum discussions. ### Tech Partner Program Full-service development for OP-native projects lacking frontend or infra capacity. Shipped apps for Pike, Namespace, Candide, Silo, and others — 1000+ hours of work. If you're building in an L2 ecosystem and need a team that can ship across the stack, this is the kind of work we do. > "See how we work" > "Bring us the problem" ---