Working session. DoubleCheck-first Phase 1 scope, live demo, and a written boundary before we hang up.
"Let's start with DoubleCheck Verified first — smaller deployment, lower risk, cleaner way to get set up before touching the RR team."— Dave, via text, 2026-04-20
What this changes. Pat is the primary champion and daily operator for Phase 1. Dave stays exec sponsor, part-time. Alexa is out of scope until Phase 2 — when a win story exists to hand her. The $3M DoubleCheck raise makes the budget path cleaner too.
DoubleCheck-first is actually stronger positioning. A 1.5-FTE team with a technical-fluent lead and a clear workflow surface (inspections) is a cleaner place to prove the pattern than a 10-person service org with an AI skeptic at the COO seat. We'd rather earn RR via a DoubleCheck win story than cold-pitch Alexa.
How this got into the brain: Dave's text on the morning of this call. Pre-call brief was rewritten to reflect it. See the artifacts ↓
Fully transparent — these are the moments we wrote down as the ones that changed how we're going into today. In your words, not ours. Verbatim from the call.
"Why didn't you start here, Will?"
"Let's assume we're going to be early adopter, like a pilot slash design partner."
"Pat is going to be your CTO, for lack of a better term. I will be the 'let's go right now, what could go wrong' — and Alexa will be 'everything could go wrong.'"
"Get rid of it. Get rid of it. Get rid of it."
"Show me, don't tell me."
"This sounds like the Jetsons, but we don't have flying cars right now because it just, like, defied gravity. We're going to need some level of show me, don't tell me — not signing up for a 12-month relationship."
Two stacks worth seeing. The first is what we made for you. The second is what we've made for ourselves using the same system — because we wouldn't sell you something we aren't running on.
Generated by /pre-call-brief. Caught the DoubleCheck pivot (after Will patched the texts in manually). Sent to our inbox.
Generated by /post-call-debrief. 5 quotable moments, stakeholder map, assessment, red flags — fully transparent.
Drafted retrospectively. "Show don't tell" + start-small framing. Sits in Will's Gmail drafts, unsent — today's call is the better version.
Five parallel research deliverables + exec brief. ~10 minutes of agent work. Deployed to Cloudflare. Dave can use it in raise conversations or a Phase 2 Alexa pitch.
→ Walkthrough in Section 04Pat's tool list, mapped against MCP availability. Five tools ready on Claude Team today; four thin wrappers; one unknown (Comparent — Pat, what is this?).
→ See Section 08Enriched automatically when the debrief ran. Alexa: journalism background, just handed off the 6-year RR newsletter — bandwidth signal. Jake: Pat's ex-NoiseAware shadow reviewer.
Weekly change digest — 5 enacted (Phoenix ADU, Decatur AL, Ocean City MD, Riverside CA, Columbia Falls MT), 3 in-committee, 2 platform updates (Airbnb ToS, VRBO fees). Manual v0.1; shows the shape.
Dave, you said you want Rent Responsibly to be one of the best AI users in the world. Here's what that looks like for us.
Color palette, typography, mark system (the asymmetric constellation at the top of this page). Auto-deployed. Same pattern we'd ship for RR's brand system if that's ever in scope.
The operational view into the brain — updates, activity, skill usage. Prototype of the "visible brain" every client deployment gets on top of the chat surface.
Every artifact you've seen was generated from this. Will's going to load it live in Section 07 and you'll see every open founder decision we're running against today.
Five real queries queued for Section 06. Dave, you saw this on 4/14 — Pat, this is the primary surface Phase 1 delivers. Everything else is scaffolding.
We wouldn't sell you something we're not running on ourselves. Everything above is in production for Atlas Mynd today.
A design-partner briefing pack, not a cold pitch. Five research deliverables + executive brief. ~10 minutes of agent work, drawing from the Atlas Mynd brain plus public sources.
Open the Intelligence Brief ↗The 18–24 month first-mover window in the alliance layer. AI adoption in your direct competitive space is structurally zero.
~142 hrs/week of AI-assistable work across your team. ~3.5 added FTE of capacity without hiring.
3,500–5,000 active local STR regimes. Your interpretation moat — the hard 90% that Quorum/FiscalNote don't touch.
Why augmentation — not hiring — makes the 50-association vision arithmetic instead of aspirational.
The DoubleCheck Inspection Knowledge Brain — Slack bot + Pat's folders. ~15 hrs/month returned; break-even on pricing by design. The Phase 1 return isn't economic. It's the pattern proof that justifies Phase 2.
Numbers are estimates with conservative bands — flagged as such throughout the brief.
Each skill is a markdown prompt — version-controlled, transferable, inspectable. Three of ours produced everything you've seen today.
Pulls from brain + web search + SEC EDGAR + LinkedIn. Caught that Dave's surname has two s's. Missed the DoubleCheck pivot because SMS isn't in the brain yet — Will patched it manually.
Processed the 4/14 transcript into a debrief, a follow-up draft, and enriched stakeholder pages (Alexa Nota, Jake). Auto-researches every new name mentioned.
Spawned five parallel research agents, composed the executive brief, deployed it to Cloudflare Pages with the --publish flag. ~10 minutes start to finish.
When you take over the brain, these skills come with it. Fork, edit, and make them yours.
Five real queries Will will ask, live, on screen-share. If one falls flat, that's real data — we flag and fix.
Action: Will tabs to Slack. Expand any query below for why-this-one rationale.
Two prompts. The first is how Will starts every Monday. The second is a test of whether the brain can push back on its own people when they're over-committing.
Claude Code follows the brain's startup protocol: reads the company CLAUDE.md, loads firm/memory.md and firm/decisions/pending.md, asks "Who's working today — Will or Justin?", then prints every open founder decision, grouped by type.
Why this is worth a minute: the brain is a working artifact, not a demo. This is Will's Monday morning — 20+ open decisions, visible on screen, ready for action.
First-principles reasoning over the brain's content plus our writing-style rules for external deliverables. Pat and Dave watch the brain help us under-commit to them — exactly the posture they told us they want.
Live theater — if the model misfires on screen, we'll walk through it together. That's also transparent.
Justin's scoping of Pat's 12-tool list, reshaped for DoubleCheck-first Phase 1.
Everything else in the 12-tool stack stays out of scope for Phase 1 — we'd rather fully resolve a narrow surface than half-wire everything.
Honest flag: Hivebrite plan tier is the single biggest technical risk for Phase 2 — API access is gated. Pat, can you confirm the plan tier before we promise a timeline?
The most important block on this page. We leave the call with this boundary in writing, or we've fallen short of what we came here to do.
Sub-account in your name, fully transferable, no SaaS lock-in. If the pilot doesn't earn Alexa's confidence, you keep the brain and we walk away. That's the whole "everything is yours" principle in one line.
Slack bot live against the scoped corpus. Working, observable, in Pat's hands.
A short retrospective in Pat's voice — the artifact Dave can hand Alexa when Phase 2 comes up.
A written Phase 2 scope doc naming exact folders + workflows + guardrails for the RR extension.
Predictable cadence. Small weekly check-ins. A retrospective that earns Phase 2 — or doesn't.