We help small teams
build and innovate
with AI.
Fractional Chief AI Officer for mid-market teams. Strategy, prototypes, and the discipline to kill the bets that won’t work. One senior operator — not a pod of juniors.
- 6-month engagements
- $5–15k/mo
- In production at CPG, B2B, Health Tech
- Cost of a CAIO
- $400k → ~$10k/mo
- Engagement
- 6 mo
- Initiative portfolio
- 3 / 7
Full-time vs. fractional
Then quarterly
Bets per quarter
The 30-second version:
- What it is
- 0→1 AI Partner (operating as a fractional CAIO)
- Who it's for
- Small teams taking on big AI initiatives
- Pricing
- $5k–$15k/mo — fixed retainer or capped days
- Engagement
- ~6 months, then quarterly
What this actually looks like.
A full-time CAIO: $400k + equity + 12 months to hire.
Strategy
- 01
Sit in on executive AI decisions.
In the room, not in a monthly report.
- 02
Kill projects that won’t work.
Most fail an honest ROI read — better to know now than in Q4.
- 03
Run the initiative portfolio.
3–7 bets, scored on effort, risk, and P&L impact.
- 04
Write the briefs your team builds against.
What “done” means, and what we won’t do this quarter.
- 05
Read vendor contracts before you sign.
Especially the ones called “just a pilot.”
Build
- 06
Build when it makes sense.
Prototype, ship, hand off.
Three things. One person.
Most people in this market sit in one circle. A few sit in two. Almost nobody sits in all three.
The rare overlap — strategist, builder, end-user obsessive — in one person.
What your senior team’s week looks like on AI.
A second mode of working. The CAIO retainer covers the company. This one covers the people running it — how the CEO, COO, CFO actually use AI in their own week.
- 01
Get the senior team set up in Claude.
Most accurate on real business work, best tone for executive writing, most defensible on data security.
- 02
Visibility, on a schedule.
Meeting notes and Slack threads pulled into scheduled tasks that surface action items and owners.
-
★ Most requested
03
A personal Chief of Staff for everyone.
Each senior leader gets their own Claude project — their context, their automations, tuned to how they decide.
-
★ Most requested
04
A design system every CEO can wield.
Brand voice, typography, layout loaded into Claude. Board decks and one-pagers without a third round back from marketing.
- 05
Audit where AI helps, and where it doesn’t.
Map each leader’s responsibilities; sort by where AI does the work, assists, or stays out.
- 06
Hire AI-native employees.
The signals worth listening for in a thirty-minute call. Not a job title — a way of thinking.
Different shape from the fractional CAIO retainer. Usually a focused stint — six to ten weeks — that ends with the senior team running on their own. Book a call →
Mid-market AI, right now.
- Your board wants an AI strategy.
- Your inbox has forty vendors offering one.
- Three of your VPs have already bought something on a company card.
Mid-market gets two or three honest swings a year. Picking the wrong ones is more expensive than missing.
Someone senior has to make the calls. That’s what we do.
Three problems, recently solved.
- CPG № 004.1
Hundreds of inbound a week, no headcount added.
A CX agent that handles email and voice, with the policies and escalations underneath that make it feel like a teammate.
Direct-to-consumer · In production Read CloseThe problem. Top Cup’s direct-to-consumer business took off. Inbound customer messages — order status, sizing, custom requests, returns — went from a manageable trickle to hundreds a week. Adding a support team would have eaten the margin the new channel just created.
What we did. Built Luna, a CX agent that handles email and voice. The interesting work wasn’t the agent — that part is close to commodity now — it was the system underneath: the policies, the edge cases, the escalation logic, and the feedback loop that made Luna feel like a teammate instead of a bot. Top Cup added the channel without adding the headcount.
- B2B № 004.2
More sales, same sales team.
Replaced “hire more reps” with better intelligence — churn signals, proactive outreach, hot prospect timing.
Sales intelligence · Engagement Read CloseThe problem. A mid-market B2B consumables company wanted to grow sales. The traditional answer — hire more reps at $120k loaded — was expensive and slow, and the reps they had were drowning in the wrong accounts.
What we did. Replaced “more reps” with better intelligence. Predictive churn signals flagged accounts about to lapse before the rep noticed. Proactive outreach nudged customers toward the next reorder window. Just-in-time sales intelligence put the right B2B prospects in front of the reps when the signal was hot, not a quarter late. The sales team didn’t get bigger. It got a better map.
- Health Tech № 004.3
12-month product cycles → 2 weeks.
Two AI systems running in parallel — one modeled the business, the other shipped working prototypes — so the team stopped guessing which version to build.
Venture-backed · Engagement Read CloseThe problem. A venture-backed health tech company was a year into a product that wasn’t finding fit. Before writing another line of feature code, they needed to know which version of the business actually worked — different pricing, different customer segments, different service models.
What we did. Two AI systems working in parallel. The first modeled the business — every permutation of pricing, cost, channel, and customer segment, ranked by unit economics. The second stood up real, working prototypes of the top-ranked versions, fast enough to put them in front of actual users. Prototype-to-test dropped from twelve months to two weeks. The team stopped guessing which version of the company to build.
Aaron.
Two decades as the person a CEO calls when a new initiative needs to get off the ground. Most of that was pre-AI — ops, product, go-to-market. The tools changed. The job didn’t.
Now the work is helping small teams figure out which AI bets are worth making, then building the first version of the one that is.
- 01Strategist
- 02Builder
- 03End-user obsessive
Stealthy Good is based in Boone, NC, and takes a small number of retainers at a time.
Notes on AI, mid-market, and the vendor-industrial complex.
Published occasionally. Written for operators, not marketers.
- 10 min
Mythos just moved the goalposts on software security.
What to do if you’re not a Fortune 500 bank — and why specialized software vendors are the ones most exposed.
- 9 min
AI at work isn’t a technology problem.
Why most rollouts are stalling — and the three things mid-market leaders should actually be focused on.
- 7 min
Most of this won’t work.
On the discipline of killing AI initiatives before they eat your quarter.
Questions you were going to ask on the call anyway.
-
What does this cost?
$5–15k/mo, fixed or capped-days. Numbers on the intro call, not a “contact sales” maze.
Two shapes. Pick the one that fits how you work:
Fixed-fee retainer. One number per month. I figure out how to add the most value against the priorities we set together. Best when the work is fluid and you trust me to run at it.
Capped-days retainer. 3 to 6 days a month, your pick. Best when your finance team wants a line they can audit, or when your scope is well-defined.
Both $5k–$15k/mo depending on scope. Both renewable quarterly after the first six months. Full pricing on the intro call — no “Contact sales for pricing” nonsense, you’ll hear a number within the first 20 minutes.
-
How long does an engagement last?
Six months, then renewed by the quarter. Most useful work happens in months 2–5.
Typical retainer is six months, renewed by the quarter after that. Most useful work happens in months 2–5. We’ll tell you when to stop; that’s part of the job.
-
What’s out of scope?
No full-time placement, no 200-person programs, no crypto.
A few specific things we don’t do:
- Full-time executive placement
- 200-person implementation programs
- Computer-vision consulting for autonomous vehicles
- Crypto
- Chatbots whose only job is to be a chatbot
If the work doesn’t change a number on your P&L within a year, we’re probably the wrong firm.
If you got this far, you probably already know whether we’d be useful.
Thirty minutes, no slides, no discovery-call theatre. Bring the problem you’re actually stuck on. We’ll tell you if we can help, and if we can’t, who might.