Playbooks

Playbooks for the AI-run business.

The systems we design and ship, written up as blueprints — what they kill, how they run, and the diagrams we'd draw on your whiteboard.

Featured · Agent systems

What it looks like when AI agents staff an entire company

We modelled a seven-department business run by 133 scoped agents under a human oversight layer — sales research, support triage, finance reconciliation, content, and the plumbing between them. The surprise isn't that it works; it's where it breaks, and why the human layer is the load-bearing wall.

7

Departments

133

Scoped agents

~$500/mo

Token spend, not payroll

Playbook 001 · Agent systems · 12 min · read the playbook →

Human oversight

approve · escalate · own the judgment calls

Sales

24 agents

Marketing

21 agents

Support

28 agents

Finance

14 agents

Ops

22 agents

HR

9 agents

Engineering

15 agents

one shared memory · one audit log · every action reversible

Voice AI · Trades

Marketing for urgent-service trades is an answering problem

When water is coming through the ceiling, nobody browses. They panic-search, call the first number, and hire whoever answers. The whole funnel is the first ring.

PANIC SEARCH"burst pipe near me" · 2:14 am
ANSWEREDvoice agent · first ring, human tone
TRIAGEDemergency? crew paged. quote? booked.
FOLLOWED UPreview request · maintenance plan offer

the ad money is wasted if nobody answers the phone

Playbook 002 · 8 min · read the playbook →

Patterns

Ten industries, one operational leak

Across every audit we run, the biggest cost hides in the same two places: friction moving data between systems, and latency in communication. The uniforms change; the leak doesn't.

Data transfer friction84%
Communication latency71%
Manual admin load66%

same disease, ten uniforms — the fix is always the boring middle

Playbook 003 · 10 min · read the playbook →

Sales systems

A sales research system that ends the 70% problem

B2B reps spend most of their week scraping, verifying and hunting decision-makers. The fix is a nightly research pipeline that hands every rep a ranked call list by morning.

SCRAPEaccounts, signals, tech stack — overnight
VERIFYcross-checked · stale data discarded
BRIEFone page per prospect, in your format
CALL LISTranked · reasons attached · 8:00 am

reps sell; the system does the 70% that wasn't selling

Playbook 004 · 9 min · read the playbook →

Regulated industries

An AI marketing department for a law firm

Legal marketing is expensive because compliance review sits on every word. The answer isn't less review — it's a pipeline where the ethics check happens before the partner ever sees a draft.

DRAFTcontent agent · SEO + ads + social
COMPLIANCEethics rules checked · citations pinned
PARTNER OKone approval · not one rewrite
PUBLISHevery channel · one voice · logged

the gate stays human — it just stops being a bottleneck

Playbook 005 · 11 min · read the playbook →

Field services

Contractors don't have a marketing problem

They have a follow-up problem, an intake problem, and no visibility into either. When the owner is on a ladder, leads go cold — no ad budget fixes that. An intake agent does.

LEADform, call or referral — owner on a ladder
INTAKEscoped by agent · photos requested
FOLLOW-UPday 1 · day 3 · day 7 — automatic
VISIBILITYevery job, every dollar, one board

not a marketing problem — a follow-up problem

Playbook 006 · 7 min · read the playbook →

Browse all

Every playbook, one click.

Want one of these diagrams
drawn for your business?