On a five-person team, leading means reviewing, unblocking and deciding what's next. I also built the tooling that lets each of us ship like a bigger team: AI agents with shared rules, and a loop from customer tickets straight to the board.
Customer issues arrived by email and chat, and AI coding help was ad hoc, with no shared rules for how the codebase works.
Tickets raised in the app become Trello cards and sync back every 5 minutes, and AI agents work from shared skills and rules that update themselves as they learn.
The problem
A small team building an enterprise product has more work than people. Context lived in heads, customer issues got lost between channels, and every engineer prompted AI tools differently.
The idea
Write things down so people and agents can share them, and connect customers to the roadmap with no human copy-paste in between.
How I did it
- Wrote the BHyve AI v4 roadmap (Jun 2026): five phases from foundation to product loops, with effort sequencing and the decisions already settled.
- Built the frontend agent kit (Apr 2026): 6 skills and 3 rules covering the API, architecture, components, data types, domain logic and styling, including a rule that the agent updates its own rules as it learns.
- Wrote about 60 design notes in five months. Each records the measurement, the decision and the trade-off, so the next person doesn’t rerun the experiment.
- Shipped in-app support tickets that become Trello cards, with retries, rate limits, environment filters and every piece of user text treated as untrusted.
- Built a personal daily triage daemon that reads the bug board each morning and drafts fix plans with an AI agent; over 61 runs it triaged 49 cards.
The impact
The team ships around 1.7 production releases a day with four code owners reviewing each other, and customer issues land on the board within minutes of being raised.
Leverage on a small team comes from writing things down well enough that both people and agents can act on them.