Launch
Presenting on
Intake forms + OpenClaw + Notion for better deal-pipeline insights.
Agentic VC Dojo — Launch Session
The next 60 minutes
Format
- 5m Framing The Dojo concept, goals, structure.
- 20m Inspiration set Four 5-min speed demos.
- 5m Polling & backchannel Live polls, discussion.
- 20m Town hall Bring a workflow you're trying to operationalize.
- 10m Q&A + next steps Follow-ups for the Dojo community.
With
Speed-demo presenters
Hosted by Michael Staton & Emily McMath
A brief moment about how this came to be.
I had AI all wrong.
It took me months of frustration to realize…
AI is not like the Jetson's robot Rosey.
AI is more like trying to tame a wild animal.
A brief moment about how this came to be.
My Kauffman experience was awash in irony.
I was like, the guy, of Education Innovation. Like, globally.
And Class 20, we were the best. But we had a lot of whiners who thought they weren't "learning" much.
After a year of hearing whines, I figured: we all know a bunch.
Why we're here
What if we just self-organizedour own sessions?
Peer Learning was born.
- Self-organizing, and matching who could teach what and who wanted to learn what.
- And, sticking around an extra half-day to make time for ourselves, focusing on whatever was most important to us.
- Weirdly, most everyone, especially the whiners, all agreed.
- And, so we did. And so did classes 21 through 23, I know, and through 25, I hear.
Enough about me...
Why we're here
A peer-learning Dojo for VCs building with AI.
- Self-organizing working groups taking on real projects.
- From "I use ChatGPT for emails" to "I'm shipping agents into our deal flow."
- Elbow grease encouraged. Casual observers welcome — but no promises.
How you participate
Bring an offer. Bring an ask.
If you can contribute
What can you offer?
- What have you figured out?
- What kind of legwork would you be willing to do?
- What rabbit holes are you willing to go down?
If you need a lift
What challenges do you have?
- Where are you stuck?
- What would amplify you in the near and long term?
- What bottleneck at your firm do you wish you could resolve?
Who's in the room today
A working room of VCs already building
- 60 Registered Kauffman Fellows + invited peers
- 37 Firms represented Distinct funds, on five continents
- 73% Already building Of survey respondents — workflows, agents, or both
- 40 Shared their context Survey responses we'll workshop together
Where the room sits today
Pre-event survey signal
40 registrants answered. Multi-select on the experience question, single-pick per tier on tools.
How they describe their AI/agentic practice
Tool stack — what they're already using
Simple Intermediate Advanced Not yet
Native LLM apps ChatGPT, Claude, Perplexity
24
9
0
7
Local AI workspaces Ollama, LM Studio, MSTY
29
0
0
11
Agent workflow builders n8n, Flowise, OpenClaw, Dynamiq
25
2
0
13
In the room
37 firms already on the registrant list
Funds, family offices, and operator-led shops — early-stage to growth, North America to MENA to LATAM to SE Asia.
- 468 Capital
- Acasia Ventures
- Aire Labs
- Alpha JWC Ventures
- Amplifica Capital
- Atlantica Ventures
- Audaz Capital
- Blueprint
- Cherry Rock Capital
- Cometa
- Corner Ventures
- Defined VC
- Determined Capital
- Domo Invest
- Domo VC
- DPI
- Feature VC
- FirstMark
- Garage Capital
- Genoa Ventures
- iMena
- Invest2Innovate
- Josi Health
- Kauffman Fellows
- KB Partners
- Last Mile Ventures
- LoftyInc
- M25 Ventures
- Medici Capital
- Mercury Fund
- MGV
- Monashees ×2
- Motley Fool Ventures ×2
- Novel GP
- Prose Ventures
- Walking VC
- Wild Basin Investments
Voices from the survey
What this room is actually working on
Quotes are anonymized. Several of these are exactly what we'll workshop on Thursday.
Automating the horrible parts of being an emerging manager. Maybe we should create a Kauffman toolset.
An agentic VC stack for diligence, fund reporting, admin and portfolio support.
End-to-end automations across sourcing, screening, follow-ups, and internal evaluation — with AI summarizing calls, extracting key insights, and keeping deal-flow data organized. The goal: less manual overhead, more consistency, a more responsive investment process.
Exploring an "agent-powered autonomous startup" — 4–5 PhD-level agents doing the research, as an experiment.
Automating the grunt work: top-of-funnel curation, deep market research, market maps.
Simple integrations across our CRM and enrichment tools to manage dealflow — plus an agentic workflow that scores each deal for prioritization.
Trying to combine various personal workflows so the whole organization can use them.
Agentic workflows for deal triaging, due diligence, and memo creation.
We've built a basic IR/fundraising engine and audit/accounting support — game-changing for us. Co-pilot enterprise for MS integration and email; Claude Team for most everything else. On a 1–10 scale, we're still a 3–4.
Voices from the survey
What this room is hoping to take home
Free-response answers to 'what do you most want to get out of this session?' — lightly edited.
Best practices for a solo GP to get 80% there with 20% effort.
Understanding output quality — that's what's lacking for me right now.
Architecture, tools, and best practices for getting the structure set up in a secure and effective way.
Would love to see what others are doing for fundraising.
Receipts
Three days. Twenty-seven thousand lines.
Sunday morning → Wednesday 7am EST. Context vigilance, a pseudo-monorepo, Claude Code.
- 27k Lines of code Astro · Svelte · TS · CSS · Markdown
- 3 Days end-to-end Sun morning → Wed 7am EST
- 9 Stack technologies See below — every layer earns its place
- 5 Changelog entries today Each one a self-contained shipped feature
Stack
- Astro 6
- Svelte 5
- TypeScript
- Tailwind v4
- Astro DB
- Turso
- Reveal.js
- GSAP
- LFM (markdown)
Best-in-class modern website
- SSG + SSR per route Per-page prerender flag
- Persistent data on Turso Five tables, one auth model
- Jumbo popdown menus Projects · Working groups · Sessions
- Three modes Light · Dark · Vibrant — all token-driven
- Disciplined design system Two-tier tokens, brand-kit + design-system pages
- Per-commit changelog discipline Five entries today alone
Interactive app capabilities
- Stacks Current · aspirational · abandoned
- User accounts via OAuth GitHub + LinkedIn, roster-gated
- Involvement-interest forms Propose a project or working group
- Live polling 5 templates · audit log · 4s interval
- Embeddable slides Reveal.js + Astro components + GSAP
- Markdown materialization DB → frontmatter on session close
- 3 years ago A whole team. A month, minimum. Probably more.
- 1 year ago 3 weeks. Infantile arguments with an AI chat. Stale context every other turn.
- Now 3 days. Context vigilance, a pseudo-monorepo, Claude Code.
What works with AI
Two disciplines, one rule.
Everything as Code
- Mermaid
- PlantUML / UML
- Graphviz / DOT
- D2
- LaTeX / KaTeX
- BPMN
Everything as Primitive Data & Content
- Markdown
- JSON
- YAML
- CSV / TSV
- XML / HTML
How we ship this fast
Astro Knots — a pseudomonorepo for the age of AI-assisted development.
Co-located projects. Shared patterns. AI-friendly docs that travel with the workspace. Eight sites. The engine behind today's three-day build.
Shared patterns
Markdown pipelines, CSS tokens, OG/SEO, theme systems. Abstract what works — without forcing premature abstraction.
Context vigilance
Specs, blueprints, prompts, reminders. Structured docs that AI assistants and humans both read. The context travels with the workspace.
Build in public, ship in private
Each site is its own git repo, independently deployable. The pseudomonorepo is a dev convenience — never a deployment requirement.
Packages from authentic need
Publish only when proven across multiple sites. First: @lossless-group/lfm — shared markdown pipeline born from building the same render logic six times.
- mpstaton-site
- hypernova-site
- cilantro-site
- twf_site
- dark-matter
- banner-site
- cogs-site
- coglet-shuffle
- fullstack-vc ← you are here
Five things to try, right now
Use FullStackVC to drive real community + knowledge exchange.
-
Can you create an account?
If you don't have a GitHub account, please do that now.
-
Can you add to your stack — or suggest something we don't have?
If suggestions aren't working, drop it in chat.
-
Can you propose a Working Group or Project?
-
Can you express interest in a Working Group or Project?
-
Do you see your data listed?
No need for everyone to report the same bug. If something isn't working, we'll get to it ASAP — this should be a meaningful tool, and we're moving fast.