team.stpierre.ai / builder brief

Collins Builder Brief

The one-page map of what St. Pierre builds, what Collins owns, and what Akash keeps. Read it once. Then work from the board.

What this isYour map of the whole build.

Why it mattersIt settles what you own.

What you doSkim it, then use the board.

Time to readAbout 4 minutes.

This page wins. When an older page fights the funnel or week-one plan here, trust this page. Two things are now stale on older pages: the scorecard is agent-built, not yours, and the $1 gate is removed.

The mission

St. Pierre helps foundation and waterproofing contractors keep a 3–6 week backlog of high-ticket jobs.

Akash owns the offer, sales, and money. Collins owns the systems that run everything else. The job is to take Akash out of the how, not add things for him to manage.

Start here

New to the build? Read the onboarding walk-through first. It takes you through the vision, your start plan, the daily rhythm, and your first call — in order.

1. Vision 2. Start plan 3. Daily rhythm 4. First call

Content lanes (1–4)

Now: made by hand and need Akash's eye.↓Target: batched from templates and set standards.

Client agents (5–8)

Now: the steps are known but done by hand.↓Target: agents run the steps and flag problems.

Scorecard (9)

Now: the cockpit exists but is busy.↓Target: one clean daily huddle view.

Review gates

Sales is Akash.AI can prep. Akash owns calls, the close, and the send.

Claims need Akash.No unproven number or claim ships.

Go-live needs review.Build on test data first. Real clients go live only after a check.

Money is locked.Never change billing, logins, or permissions without a clear yes.

System work is yours.Move when it is logged, tested, reversible, and on the scoreboard.

Four surfaces are off-limits.The SMS worker, the Meta ad account, the launcher, and the scorecard tables.

What you do next

Read the onboarding walk-through, then open the builder board.
Work one lane at a time. Report by scorecard: what moved, what is blocked, who owns next, what should stop.
This week: run and start to automate the daily production machine. Ship the full slate five days straight.

Extra detail — open only when you need it

The funnel — the shape to memorize

How a stranger becomes a sales call. The $1 gate and the standalone VSL page are gone. The funnel is booking-gated (locked 2026-07-12): nobody sees a demo until they book.

Step	What happens	Owner
1. Instagram feed	Content and the VSL run as posts and ads. They answer questions before anyone talks to Akash. There is no standalone VSL page.	Akash / agents
2. Lead form / Typeform	The only two places traffic goes. Nothing else gets traffic. (Typeform = our short application form.)	Agents
3. Booking (Calendly)	The form routes straight to the calendar. No thank-you page detour. The lead books their own call.	Agents
4. Custom demo	The moment a call is booked, a personal demo page auto-builds in under a second. The confirmation text delivers the link. Automating and hardening this engine is Collins' build.	Collins
5. Sales call	Akash closes.	Akash

Your week-one build — full detail

Refocused 2026-07-13. Collins starts by running the daily production machine (newsletter, organic, AI-UGC video, carousel, tweet/note statics, daily ad flow) from the existing templates, then starts to automate the running — plus automating the sales system around the funnel. Week-1 proof = five straight days of the full slate shipped, plus a hand-back on what to automate first. The custom demo engine proof-of-life below moves to follow-on (Akash confirms on the kickoff call).

Follow-on (was week-1 per Jul 10): For 5 real prospects (named at kickoff): one link each that stacks their live ROI page with three personal audits — Brand Audit, AI-Readiness Audit, Bottleneck Audit — made from a repeatable template.

What	Detail
Friday test (done =)	Akash opens one link and would send it to that contractor the same day. Scaling to the full warm list after that is mechanical, not creative.
The pattern you extend	The ROI Calculator is live. The go.stpierre.ai worker writes each applicant to the database, and /nextsteps shows their numbers from one record. Your audits bolt onto the same one-record → one-page pattern.
Inputs you get at kickoff	(1) approved 6-asset scope, (2) the 5 named prospects, (3) a market-personalized launch plan, not a clone of a real site, (4) a claims process: you draft the claims bank Monday, Akash approves async. Unverified numbers never ship.
Secondary item	A read-only map of what tools and automations exist and what is missing. It teaches you the stack.

Not yours: the weekly scorecard (agent-built — if the board shows a scorecard card for you, flag it, don't build it) and founder talking-head content (parked).

The nine lanes — now vs target, lane by lane

Lane	Now	Target	First useful output
Static	Direction exists; production depends on Akash taste.	Weekly batch ships from standards and examples.	10 approved static concepts with hook labels.
Motion	A lane in concept, not a template set.	Motion versions come from the same idea bank as statics.	3 reusable motion templates with export rules.
Carousel	Role is clear; the workflow is not.	One idea becomes a clean hook → proof → ask.	Carousel structure and 3 examples.
YAP	Akash has raw thinking; outputs are manual.	One recording makes clips, hooks, captions, and briefs.	Transcript-to-output checklist.
Onboarding	Path is set but still needs the owner.	A paid client reaches launch prep without Akash chasing.	Missing-info tracker and a test client run.
Fulfillment	Launch checks are known but not packaged.	Each client has campaign, creative, tracking, and next action visible.	24/48/72-hour QA loop.
Finance	Runway pressure is visible; the rhythm is not simple.	Cash and payment risk surface before an emergency.	Weekly finance view.
Reporting	The idea exists; client proof needs structure.	Reports show proof, risk, and next actions with no manual assembly.	Red/yellow/green report template.
Scorecard	Cockpit exists; the huddle view needs sharper lanes.	A daily huddle reviews nine lanes, not a task dump.	One-page huddle scorecard.

What St. Pierre is becoming

AI agents do the repeatable work: drafts, data pulls, checks, routing, scoring, and reports.
Builders manage the agents: prompts, workflows, data joins, QA, rollback, dashboards, and docs.
Humans own trust: sales calls, client recovery, resets, money calls, and final judgment.
The company runs by scorecard. Green means humans stay out of the work. Red means the owner fixes the system that missed.

The build order

The phase order (Jul 10): custom demo engine → demo at scale → production and automation.

Phase	Builder output	Open when needed
Week 1 (refocused Jul 13)	Run and start to automate the daily production machine, and automate the sales system around the funnel. Proof: five straight days of the full slate, plus a hand-back. The custom demo engine proof-of-life moves to follow-on.	Start plan
Phase 1 — demo at scale	Harden the template so the demo is the standard pre-booking step for the whole warm list.	Builder board
Phase 1 — delivery	Client onboarding automation, from cash collected to launch prep.	Reference shelf below
Phase 2	Outbound packets and destination tests after the first system works.	Reference shelf below
Phase 3	No-founder production, reporting loops, and media-buying support.	Reference shelf below

Review gates — full detail

Gate	Owner	Rule
Sales relationship	Akash	AI can prep. Akash owns calls, the close, and any send as Akash.
Claims and public proof	Akash	No unsupported claim ships. Early creative needs a taste and claim review.
Live client go-live	Akash + builder	Build against test data first. Real clients go live only after review.
Money, access, permissions	Akash	Do not change money, logins, billing, or permissions without a clear yes.
Reversible system work	Builder	Can move when it is logged, tested, reversible, and visible on the scoreboard.
Locked surfaces (Jul 10)	Agent lanes	Four systems are off-limits this phase: the SMS worker (a live compliance issue this week), the Meta ad account (launched ad sets are frozen; new creative = new packs), the command-center launcher, and the scorecard tables. If your work touches one, say so in your hand-back — don't touch it.
Working rhythm	Builder	One task at a time. Every hand-back: what changed, proof it works, blockers, and the next move. Blocked over 30 minutes: send a Loom. Batch your questions. Draft, never send.

Reference shelf — nothing here is needed in week 1

What do we sell?Open when you need the offer, pitch, or objections. 6-week plan (Gantt)Open when you want the full phase order and dates. Custom Audit Engine SOWOpen when you build the demo or audit lane. Onboarding Automation SOWOpen when you build the onboarding agent. Creative Engine SOWOpen when you build the content lanes. Pipeline mapOpen when you want to see the whole business flow. Creative ops flowOpen when you want the creative request path. Marketing / ad testingOpen when you want how ads get tested and scaled. What systems already exist?Open when you want the automation map. Ad rulesOpen when you touch Meta naming, geo, or QA. GHL plumbingGHL = our CRM. Open for lead tracking and onboarding infra. Performance dataOpen when you need campaign metrics and launch QA. What is broken or drifting?Open when something looks off in the systems.