00 — Playbook

Twelve patterns.
Three tiers.
Which one is yours?

These are the shapes of work that actually land for $1M–$50M businesses adopting AI. Some we lived at Tumble. Some we see working in the market. Some are the ones we're watching emerge. None of them are theories.

We publish this because the first-mover window is narrow, the cost of being wrong is high, and most of the advice on the internet is written by people who've never shipped any of it. If one of the twelve looks like your problem, the first call is always free.


01 — Why now

The window is narrower than it looks.

The interesting number isn't how many companies are using AI — it's how much faster the ones who are are growing. The gap is already visible in the aggregate data, and it's compounding.

SMBs using AI

91%


report revenue increases from their AI work.

Revenue per employee

27%


annual growth at highly AI-exposed firms — roughly 3× the rate of laggards.

By 2027

93%


of SMBs are projected to use AI in at least one function.

Aggregate industry figures, not Clarified outcomes. The point is the direction, not the decimal.


02 — Tier one

Highest impact · Fastest to deploy

Four patterns that pay for themselves first.

01

Premium AI concierge

A paid concierge layer that handles intake, retrieves policy and product knowledge, runs diagnostics, executes standard procedures, and escalates edge cases to a small human team.

Best fit

B2B SaaS, niche software, agencies, high-LTV ecommerce

Industry shape

MVP in 4–6 weeks. Production-grade in 8–12.

02

AI prospecting & meeting-setting

An always-on outbound engine. Researcher, sequence writer, reply handler, and scheduler working as a pipeline — with humans in the loop on the replies that matter.

Best fit

B2B services, agencies, MSPs, niche manufacturers

Industry shape

MVP in 3–5 weeks. Reliable reply-handling in 8–12.

03

Agentic quote desk

Agents gather requirements conversationally, map them to the catalog and pricing rules, generate first-pass quotes, and route exceptions to humans. Humans review the unusual, not the routine.

Best fit

Distributors, manufacturers, MSPs, construction, commercial services

Industry shape

MVP in 6–10 weeks on the top 20 products. CPQ-lite in 3–5 months.

04

New-market launch kit

A repeatable motion for entering a new geography, channel, or vertical in 30–60 days — localized listings, pricing and tax readiness, compliance checklists, messaging, and ongoing optimization.

Best fit

Ecommerce, DTC brands, B2B manufacturers, software companies localizing

Industry shape

MVP in 6–10 weeks for one channel, one market.


03 — Tier two

High impact · Medium complexity

Four patterns that compound once the foundation is there.

01

Managed content studio

Agents generate campaign plans, produce multi-format assets, schedule distribution, and monitor performance. Humans keep the creative direction and brand standards.

Best fit

Marketing agencies, ecommerce, B2B services, multi-location businesses

Industry shape

MVP in 4–6 weeks. Agency-grade operations in 8–16.

02

Field-service triage & remote assist

An AI-first booking, triage, and remote-assist layer for field-service businesses. Agents answer calls, identify issues, book jobs, upsell maintenance, and sometimes resolve remotely without a truck roll.

Best fit

HVAC, plumbing, electrical, commercial maintenance, IT onsite

Industry shape

MVP in 6–10 weeks. Mature dispatch in 4–8 months.

03

Customer training & certification

An AI tutor that teaches customers how to use the product, prepares them for certification, and coaches them in context. Tutor agent plus a practice agent plus an assessment agent.

Best fit

Software vendors, industrial equipment, professional services

Industry shape

MVP in 8–12 weeks. Scaled multi-course in 6–12 months.

04

Vertical virtual analyst

A paid insights product where an agent answers questions about operational data, generates weekly insights, and benchmarks against peers. Translates business questions into governed queries with traceable answers.

Best fit

Agencies, MSPs, franchisors, B2B SaaS with cross-customer metrics

Industry shape

MVP in 8–14 weeks. Multi-tenant analytics in 6–12 months.


04 — Tier three

Higher complexity · Higher ceiling

Four patterns for the businesses ready to build something platform-shaped.

01

Personalized product line

A premium design-to-order product line where agents collect preferences, propose designs, generate variants, and route fulfillment. Mass customization without the mass labor.

Best fit

Apparel, gifting, home decor, specialty foods, subscription boxes

Industry shape

Bigger builds. Usually paired with a first capability under another pattern.

02

Managed community & knowledge hub

An agent-run knowledge hub that answers questions, routes discussions, suggests resources, and escalates to internal experts. Human moderators handle edge cases and relationships.

Best fit

Niche software, professional associations, B2B platforms, education

Industry shape

MVP in 6–10 weeks. Mature community operations in 3–6 months.

03

White-label vertical agents

Productize industry-specific agent workflows and sell them as recurring subscriptions to other SMBs — a packaged agent plus integrations plus ongoing tuning. Agent ops as a service.

Best fit

Digital agencies, MSPs, niche consultancies, vertical SaaS implementers

Industry shape

Platform build. Months, not weeks. Margin compounds with template reuse.

04

Agent-operated niche marketplace

A focused marketplace where agents handle onboarding, matching, screening, and transaction coordination. A small founding team runs what used to require heavy ops staffing.

Best fit

Specialized services, procurement networks, B2B staffing, high-friction categories

Industry shape

Bigger builds. Usually the most ambitious line of work we take on.


05 — How to tell

One number tells you if it's working.

There is a trap where revenue goes up because human labor went up, not because AI took on more of the work. The warning signs are flat or declining gross margin, flat revenue per employee, and delivery still heavily dependent on human hours.

The cleanest single metric is revenue per employee. If it's going up over time, the AI is earning its keep. If it's flat while revenue grows, the business is scaling human labor wearing an AI coat — and someone should say so.

That someone is us. We will tell you if what we built isn't moving the number, and we will tell you what to do about it.

Not sure which one

Which pattern is yours?

That's what the first call is for. Most operators come in with a problem, not a pattern — and one of the first things we'll do together is figure out which of these twelve (if any) fits what you're actually trying to solve.