The thesis
SYNRGY is a roll-up: acquire regional youth-sports operators, aggregate them onto one platform, improve operations, and build toward an exit. It runs a structured M&A and capital-formation operation with several club targets in diligence at once. That strategy — the deal thesis, the targets, the capital formation — belongs to the investor who runs the firm, not to me. My remit is operational and downstream of it: the bet I'm executing is that a small investment shop can run like a much bigger one if the operational layer is automated to the hilt — without ever letting automation touch the things that require human judgment.
What I do here
I build and run SYNRGY's AI operations layer — the automation underneath the firm, not the investment side of it. I architected a headless automation platform, replaced a paid CRM with a custom one, built the deal dashboards and investor sites, and led the migration of that system off a single Mac Mini onto cloud infrastructure with proper tunneling and sync. It's an operating engagement, not something Alpicat owns: the strategy and the deals are the investor's; the operating system around them is mine.
What's built
- ◦~23 scheduled AI skills running headless on a timer: meeting prep, call-notes processing (transcript → action items + contact intel), weekly summaries, Sunday pipeline audits, Slack-driven approval loops, target research, an industry-news digest, email-compliance scans, and a self-improving prompt-review loop.
- ◦A dual-mode architecture — scheduled headless agents for recurring ops, plus interactive AI sessions for polished, investor-facing diligence documents, sharing one filesystem.
- ◦A custom CRM — a self-hosted FastAPI + SQLite system that replaced an off-the-shelf tool, with three pipelines: capital formation, M&A targets, and ventures.
- ◦An apps layer — a due-diligence dashboard (a 95+-item DD tracker with its own API), investor project sites, and an interactive system-architecture map, auto-deployed.
- ◦A hard human-in-the-loop gate — a data-room workflow that physically blocks any automated upload to a live investor data room; everything stages for human approval first. The most important automation is the one that refuses to act.
- ◦A self-building research wiki — a knowledge base that compiles itself nightly from new material.
- ◦An AI writing-style guide — reverse-engineered from real sent email, with explicit anti-"AI-tell" rules, so anything drafted on the principal's behalf reads as authentically human.
Where it stands
Mature and in active daily operation, mid-migration to cloud infrastructure, multiple targets in diligence. It's also heading toward a product: the skills layer is being packaged so other investment shops and family offices could run the same back office.
Nearby stars
Blueprint is its lead-sourcing arm — shared pipelines, shared club database, shared enrichment tooling, one operator. Both live in the Alpicat brain collection as engagements rather than holdings.