Date: 2026-10-02 Time: 10:30 AM EDT Attendees: Michael Houck (Rye Valley), Emil Barr (Candid Network) Granola: https://notes.granola.ai/d/351c0a04-4fcc-43aa-9829-03db11ac5a69
Note: equity and retainer terms are in Owner tier
Summary
Michael and Emil aligned on Emil joining a new government-data venture (working name: Candid Network) as a co-founder-level partner. They resolved the equity and cash structure, mapped out a data-acquisition strategy, and agreed on go-to-market via direct outreach to frontier AI labs. Ohio is the first target state. They will move immediately on public records request infrastructure and a validation pilot before incorporating formally.
Decisions
- Emil joins the venture at co-founder-level equity, with a monthly cash retainer invoiced through Rye Valley starting November 1. See restricted file for figures.
- Delaware C corp setup deferred until pilot proves data acquisition is possible and buyers exist.
- Go-to-market targets frontier AI labs directly (OpenAI, Anthropic, Google), not brokers.
- Ohio first: home-state agency relationships give Michael and Will a faster yes/no.
- Data acquisition strategy: split broad public records requests into narrow per-section sub-requests, each from a separate mailbox, to force agency compliance.
- Compliance escalation: carrot (agreement on data-use restrictions) plus stick (Quinn Emanuel litigation threat). Neither approach is without complications; details in transcript.
- Pilot goal: small dataset (one state, approx. $200-300K worth) to prove buyer appetite before scaling.
- MOU to be drafted this weekend covering three-way equity splits; Will conversation happening tomorrow morning first.
- Retainer billing: invoiced to Rye Valley, credit-card auto-pay preferred over net-30.
Action items
- Michael: draft and send MOU covering equity splits for all three parties (target: this weekend, pending Will sync tomorrow)
- Emil: follow up with Ali (MicroOne) on the intro Michael sent; frame as collaborative, not competitive; schedule call with Hunter too if possible
- Michael: reach out to OpenAI contact (VP Partnerships, former investor) and Anthropic contact (technical staff) to start data format/buyer appetite conversations
- Emil: research government search benchmarks by state (what people search / get AI answers for re: government topics) to inform pricing tiers
- Michael: set up $10K/month Rye Valley invoice for Emil; first invoice covers October; auto-pay via credit card
- Will: finish detailed public records request memo and submit a large batch in Ohio in one coordinated day; use a lawyer for the calls, not founders directly
- Emil: spin up a Hermes instance for the email-writing agent and internal tooling UI (after Will’s memo is ready; no hard blockers after that)
- Michael/Emil: if Ali + Hunter are available, do MicroOne call this weekend; Emil pings first
Notable transcript excerpts
On why 25% equity beats 15% + commission:
“I would rather just give you like 25% with the understanding that like there’s very minimal expenses on this business if I’m going to be doing quarterly distributions… I just want to give you more because I think then it’s like, it truly feels like you’re a co-founder.”
On the data-acquisition mechanic:
“You can just take what would be like one big request, which is like, hey, we want all of the emails corresponding to constituent questions about the tax code - they would throw that [out] because it’s all emails about the tax code. But if we said, hey, we’d like emails specifically on what constituents corresponded around Section 11 of the tax code, and there’s like 50 sections. They have to honor that request because it’s fair enough. So you just have to take the request and submit 50 individual public records requests.”
On why direct AI lab sales beats brokers:
“The middlemen are just going to take a cut. We have the relationships to [go direct]. We should just go straight to the big frontier model builders to give them a unique data source.”
On Emil’s role (CTO + Silicon Valley sales):
“I can be CTO, I can be Silicon Valley sales guy, I can be both… I can build those [data-cleaning harnesses]. I did as a data scientist as well when I was at Uber.”
On the 10-state exit thesis:
“If we get 10 states at $2.5 million per state, that's $25 million. And sell this thing for… at least 1.5x revenue. And you’re selling directly to these labs and saying, hey, none of these other labs will be able to get this data.”
Full transcript: see Granola link above.