Blake Becker
AI Systems
Two systems I built because I needed them and still run every day. Farai governs the agents, and Career Command Center is the first venture running underneath it.
One screen for running every venture · 33 agent roles
An all-in-one control room for a team of 33 AI agents that run my ventures. Every agent is a written contract with its own authority level, token budget, turn ceiling and escalation target, and a separate governance tier audits the rest and reports to me instead of to the commander it is auditing, so no agent can quietly widen its own permissions. I talk to one point of contact, a chief of staff that knows what every seat does and hands each job to the right one. The cockpit is a single screen showing everything running in the background and everything waiting on my approval, and my calendar and to-do list sit on it too. The roster is also drawn as a town where each function tier is its own district and every agent is a character with its own look and job, so I can see who is working rather than read a log. It tracks how each seat performs, refines its own playbooks from that, and projects the whole knowledge base into Obsidian.
Built with Claude Agent SDK · Next.js · TypeScript · SQLite and Cloudflare D1 · three.js with React Three Fiber · an in-repo MCP server · Obsidian
Optimizes the odds of landing a role · 84 opportunities tracked
Recruiting software built to raise the analytical odds of landing a role, by running four things as one system that normally sit in four separate tools. It sources opportunities across several industries at once and pushes them back to you instead of making you go looking. It holds every person you have talked to along with the context and who else they are connected to, so for a given role it can name the relationship most likely to get you in the door and put that person on an outreach sequence that keeps working until it converts into a referral, a coffee chat, an interview or more. It measures what you already know against what that specific role requires and teaches you the gap on a schedule that finishes before the first interview. Every deadline, interview and coffee chat from all of it lands on your Google Calendar, so the sourcing, the relationship, the preparation and the timing move together rather than being reconciled by hand.
Built with OpenAI Codex first, then migrated to Claude Code · Next.js · TypeScript · Drizzle ORM · Cloudflare Workers and D1 · Google OAuth
Open the desktop · bhbecker@umich.edu