Engagement · Tier 3
Growth AI Accelerator
Six weeks. Three workflows in production. Measured before-and-after. Your team owns what we build — we don’t ship anything they can’t maintain. Priced $997–$2,997 based on team size and domain complexity.
What you leave with
- Three production workflows running in your stack at the end of week six — not prototypes, not demos.
- Before-and-after metrics on each workflow: hours saved, error rate, customer-facing SLA, or revenue impact.
- A runbook for each workflow your team owns and can modify without us.
- A 30-day observation plan so you’ll catch drift, hallucination, or cost creep before it compounds.
- The internal champion on your team leveled up enough to scope workflow #4 without our help.
Six-week schedule
The cadence is engineered around OODA loops — observe, orient, decide, act — at one-week intervals. Every week ends with a shipped or killed decision.
Week 1
Diagnose
Full workflow audit. We ride along with the team for two days, map current state, and apply Theory of Constraints to find which three workflows to attack. We pick the three; you approve or swap.
Week 2
Design (Workshop 1)
3-hour working session. We design all three workflows end-to-end on paper — tool selection, prompt architecture, evaluation criteria, failure modes, rollback plan. Nothing is built yet.
Week 3
Build workflow one
Your team implements under our direction. We pair on prompts, orchestration (n8n, Zapier AI, or custom), and the evaluation harness. Ship by Friday or we restructure.
Week 4
Build workflow two (Workshop 2)
Mid-engagement workshop uses workflow one’s real data to recalibrate design choices on the remaining two. Second workflow ships end of week.
Week 5
Build workflow three
Last workflow. By now your team is driving and we’re reviewing. We spend the reclaimed bandwidth on the observation plan and cost model.
Week 6
Measure (Workshop 3)
Final workshop: we walk through 14 days of production data, compare to baseline, and decide what to keep, kill, or expand. You leave with the measurement runbook and the plan for the next three workflows.
What’s included
- 3 live workshops (3 hours each, on-site or Zoom) — weeks 2, 4, and 6.
- Weekly 60-minute working sessions in the off-weeks.
- Shared Slack channel for the full six weeks and 30 days after delivery.
- Hands-on implementation pairing — we write prompts, design evaluation harnesses, and debug orchestration with your team, not for them.
- All runbooks, prompt libraries, evaluation suites, and measurement dashboards yours to keep.
- One 60-minute revisit call at day 60 to pressure-test what’s still running.
Questions
Why three workflows specifically?+
Two is not enough for your team to internalize the pattern; four starts splitting our attention and dropping quality. Three is the load that keeps senior attention on each workflow every week.
What if our current stack isn’t ready?+
The diagnose week surfaces that. If we hit infrastructure that blocks all three workflows, we stop, refund the remaining weeks, and write you a foundation playbook instead. Has happened twice in 30+ engagements.
Do you write code?+
We pair with your engineers. We write prompts, evaluation harnesses, and agent scaffolding in Claude/GPT-5/Cursor during live sessions. We don’t ship production code solo — that model creates dependency we refuse to build.
Why the price range?+
Anchor at $1,997 for a team of 3-10. The $997 floor is a founder-led build with one implementer. The $2,997 ceiling is for teams of 20+ with multi-tenant data complexity, compliance review, or regulated-domain testing requirements.
What if we miss a workshop?+
Workshops are recorded, but the engagement is designed so two of them are load-bearing (weeks 2 and 6). Miss one of those and we reschedule — we don’t compress the timeline.
What models and tools do you actually use?+
Depends on the workflow. Claude for reasoning-heavy orchestration and long-context reads. GPT-5 for multimodal and voice. Cursor for engineering work. n8n for self-hosted orchestration, Zapier AI for teams that already live in Zapier. We stay current and pick per-job, not per-relationship.
Scope an accelerator
We run three accelerators in parallel at most. Book a 30-minute intro to confirm fit, scope, and a start date within the next 4-6 weeks.
Book scoping call