07 / AI Automation · NEWAI features and agents, embedded in custom apps you own.
AI work splits into two kinds of vendors: specialists who only do AI but can't ship the surrounding application, and full-stack developers who can ship the application but treat the AI layer as someone else's problem. I'm the second kind, with the AI layer included — useful when you want one person to build the whole thing and the AI is part of it, not a separate vendor relationship.
Every project is custom-coded against the model APIs (Anthropic, OpenAI, etc.) — not built on a no-code platform you'll pay rent on forever. Your prompts, your data flow, your agent logic, all running on infrastructure you own.
AI features in web apps
Chat assistants, content generators, summarization, semantic search inside your custom app.
Workflow automation
Auto-classify inbound leads, draft replies, transcribe and summarize calls, process documents.
Legal-tech automation
Contract review, NDA generation, intake forms with structured extraction. Built by a JD.
Custom agents
Multi-step agents using Claude / OpenAI SDKs, wired to your business logic and data.
- Discovery call to scope the use case (free first 30 min)
- Custom prompts, evals, and guardrails written for your data — not generic boilerplate
- Server-side API integration so prompts and keys aren't exposed in client code
- Cost monitoring + rate limiting baked in (so an unexpected spike doesn't blow your bill)
- Honest framing: I'll tell you when an AI tool is the wrong answer for your use case