Deploy agents into your firm, not around it.
An agent stack and a second brain for emerging venture firms — installed straight into the Claude and GPT you already run. Operating in days, not quarters.
You invest in AI-native companies. Is your firm one?
You back founders who build with agents from day one. But inside the firm, the work still runs on memory, scattered docs, and a partner's inbox. The tools your portfolio takes for granted haven't reached your own desk.
VCOS closes that gap — without a platform migration or a six-month rollout. Make the most of your management fee by doing more with less.
Three pieces, one repo, zero lock-in.
VCOS installs into the assistant your firm already runs and gives your agents a memory that stays in sync. Everything lives in a git repo you own.
A firm-specific git repo
We stand up a private repo — vcos/your-firm — with the skills, prompts and workflows your firm needs, tuned to your thesis. It drops straight into Claude or GPT.
No technical skills needed. If you can use Claude, you can use VCOS — we handle the git side, you just work in the assistant.
A second brain, kept in sync
A pre-built knowledge vault — deal notes, founders, theses, memos — created and maintained by the agents and synced across your team, so context is always current.
Inside the Claude / GPT ecosystem
You inherit marketplace connectors and enterprise-grade security from day one. No middleware to babysit — just the assistant your team already opens every morning.
A minimum that's actually valuable.
We've done the work to build a stack worth running on day one — customizable to emerging and small VC firms, and ready to grow into custom agents when you are.
The agent stack
A curated set of skills, prompts and workflows that drop into the assistant your firm already runs.
- Deal screening against your thesis
- IC memos drafted from your notes
- Founder & market research on demand
The second brain
An Obsidian vault of deal notes, founders and theses — interlinked, queryable, and maintained by the agents.
- Synced across your whole team
- Smart retrieval for sharper context
- Your firm's memory, queryable
No lock-in, ever
It all lives in a git repo you own. Take it anywhere, update it over time, or use it as the base for a self-hosted stack.
- Markdown in, markdown out
- Swap models or self-host anytime
- A clean path to mid-market tools
Your firm's memory, queryable.
Every note, deal and thesis interlinked in one vault. The agents don't re-read everything — they retrieve only what's relevant, so context is sharper and tokens are cheaper.
Smart retrieval, not brute force
The vault's links and structure let agents pull the few notes that matter — not the whole archive.
Lower token cost, every call
Less context shipped per request means meaningfully lower spend — and faster answers.
Synced across the team
Maintained by the agents and shared by everyone — the firm thinks with one memory, not five inboxes.
Bring your own tools. Secure from day one.
Your agents work with the tools your firm already runs — connected through the Anthropic and OpenAI marketplaces, out the gate. You inherit enterprise-grade security from the assistant. For anything not supported natively, we build the connector.
Marketplace-native
Connect through the Claude and GPT marketplaces — no middleware to stand up or babysit.
Secure from day one
Enterprise-grade security inherited from the assistant — not bolted on after the fact.
We fill the gaps
Not supported natively? We build the connector — and suggest the data tools worth adding.
The window is open. Step through it.
Token economics are unusually favorable right now — and VCOS is built so you keep the upside when they change.
Ride the subsidy while it lasts
Run on subsidized usage through Anthropic or Codex Max plans and get real agent value at a fraction of the eventual cost.
No regret when prices shift
Because the stack is a portable repo, you can swap to cheaper models, or self-host, the moment the subsidies fade. No rebuild.
A base you grow into
Begin with a valuable minimum, then expand into custom agents and mid-market solutions on your own timeline — same foundation.
One setup, your cadence.
A fixed setup gets you live. A light retainer keeps the stack current as the technology — and your thesis — move.
- Private vcos/your-firm repo
- Curated agents & skills
- Pre-built Obsidian vault
- Installed into Claude or GPT
- 3 months of tuning & refinement included
- Regular usage & agent review
- Training on the latest advancements
- Customization & new agent creation
- Dedicated Slack support channel
- Everything in Retainer
- All git pushes & maintenance run by us
- Embedded, hands-on partner
- Priority roadmap input
Setup is one-time. Retainer is monthly or quarterly — same offering, your cadence.
The questions partners ask.
Run an AI-native venture firm.
Book a call and we'll scope a VCOS stack tuned to your thesis — operating in days, not quarters.