Experienced engineers, AI-forward.
We build AI applications for early-stage teams. Senior leads, parallel coding agents, production from day one.
Before Code Corp
Our leads built and ran the platforms behind Casper, Obama for America, Hillary for America, HubSpot, Flexport, Optimizely, GoodRx, and The New Yorker. Code Corp is what they are building now.
Two things,
same engineers.
Most of our customers buy both. The same senior engineers run both engagements.
App Development
Custom apps and the AI behind them. Web, iOS, Android.
- Greenfield products, rebuilds, feature pushes
- Production AI: agents, RAG, knowledge graphs, evals, observability
- Senior lead reviews every PR, on the call with you, and on the call with the agents
Claude Deployment
Deploy Claude across your product and engineering teams, then connect it to the systems we wish came with an MCP server already.
- Org-wide rollout, training, internal champions
- Custom MCP servers for in-house platforms (the off-the-shelf integrations stop where your stack starts)
- Strongest in product, engineering, and ops
How the work
actually gets done.
One senior per agent team.
Kyle, David, or Mohammed leads every engagement. They scope the work, review every PR, and stay on the call with you. Every engineer on the team was personally recruited and trained by Kyle. No agency hires. No contractors.
AI-native delivery.
Each engineer runs up to eight Cursor agents in parallel. The agents do the typing. The senior reviews every PR. Project-specific Cursor rules, planning templates, TDD enforcement, frozen-golden evals for AI features. Sentry, LangSmith, structured logs from day one.
The numbers,
with sources.
Internal tracking across active engagements, measured against the same teams' pre-agent baselines.
GitHub analytics. The other 15% is the part that needed a senior engineer.
Sentry, same teams, same products, before and after coding agents went in.
One peak week on a six-person team, eight parallel agents. We publish the inference number on purpose.
What we have
actually built.
Anonymized where we have to be. Real numbers, real failures, the parts we'd do differently.
Home management platform (anonymized)
We embedded a six-person team running parallel coding agents and shipped 89% faster against the team's pre-agent baseline, while building the multi-agent backbone (RAG, knowledge graphs, document understanding) the product runs on today.
Loan servicing client (anonymized)
Three-phase build: a document understanding pipeline that extracts and validates loan data, a natural-language retrieval assistant for internal staff, and the recurring reporting and email workflows that used to live in someone's calendar.
Consumer brand (anonymized)
Technical design and ground-up build of a data-intensive web app on Next.js, Bun, Postgres, and AWS. Architecture chosen for one reason: 90% of the work would be done by coding agents under senior review.
Blog.
Book Review: Prompt Engineering for LLMs (O'Reilly early release)
This new book on prompt engineering has a lot of useful tips and tricks. I review some of my favorite parts.
Oct 09, 2024 · 5 minArchitecture · Engineering · MethodologyPremature optimization can sink your startup
For successful outcomes, it's important to narrow the initial problem solving scope. Focus on the customer problem and maintainable software.
Oct 07, 2024 · 3 minAgile · Methodology · EngineeringFail Fast, Learn Faster: How to Ship Value Without Wasting Time
Don't waste months building the wrong thing. This guide shows you how to test, learn, and adapt at lightning speed, helping you get customer feedback fast and avoid costly mistakes. Time is your most valuable resource—use it wisely!
Sep 18, 2024 · 9 minGot something
to build?
Tell us what you are trying to ship and what is in your way. We will tell you whether we have built something like it before, and whether we are the right team for it. Same business day.