Andrei Matveev. Playbook · 2026-09-03Book an intro call
Six steps · for marketing and GTM teams building with AI

The GTM AI harness playbook

Engineers describe an AI agent as a model plus a harness. The model can be swapped for another vendor's, while the harness stays.

This playbook is for growing marketing and GTM teams who want to go AI-native, and it exists to help them skip the problems that catch most teams a few months in.

For many marketing teams, their AI-native GTM journey begins with building an AI skills library. Each skill serves as a written instruction for one job executed by AI. The most-installed marketing skill library on GitHub is past 280,000 installs, so plenty of teams have already taken the first steps.

The trouble starts a few months later, and it tends to look like this.

What goes wrong a few months in.
Generic content. Whatever nobody wrote down, the AI fills in with the industry average. So the content reads like every other company's, and your senior people re-edit it by hand and spend endless cycles iterating with AI.
Output that changes from run to run. Ten runs of the same job read your strategy ten different ways.
Skill copies that drift. Team members copy, then edit skills, and each copy carries its own version of the truth.
Claims nobody cleared. A client name or a claim that no one approved appears in AI output, because nothing was proactively checking.
Files nobody owns. The files used by the AI go out of date, and there is no formal process for updating them.
Every one of these is a missing part of the harness, and a better skill or a better model fixes none of them.

A skill should only define how a specific job runs. The full GTM AI harness is everything else needed to make AI outputs safe to publish under the brand's name.

The playbook below is about building those parts around the skills. Andrei Matveev wrote it, and it combines what he has learned building harnesses for clients with what teams like Zapier's have published about running their go-to-market on AI. It starts with what a harness is made of, then walks through the six steps that build one, and it ends with what a harness can't do and where to start.

The destination

Five parts, and most teams build only one

An agent has five parts. One is the model, which can be swapped. The other four are the harness. Follow one asset, say a blog post or a customer deck, and it passes through all five on its way from the request to the approved version.

The five parts of an agent, from the request to the approved asset.
A request comes in
ContextThe files the AI reads before it writes anything. They hold what is true about the firm, and named people check them and keep them up to date.
SkillsWritten instructions for how each job runs.Where most teams stop
The modelThe AI itself, from whichever vendor you pick. Swap it for another and the harness stays.
ChecksAutomatic checks on every draft, then a named person's sign-off.
An approved asset goes out
Upkeep, the fifth part. Corrections that repeat become standing rules and test cases, so the next asset starts from them.

Notice where the skills sit. They are one part in the middle, and the parts before and after them are what let the asset go out under the firm's name. The playbook below builds those other parts one at a time.

The playbook

Six steps, in the order they are built

The harness is built in this order, because each step builds on the ones before it. Every row links to its section, and every step ends with what changes for the team once it is done.

Pull the truth out of the skills

What's true about the firm should live outside the skills, in one shared place that skills can read. Keeping the truth apart from the skills is the first thing to build.

Take the truth out of the skill, and keep it in one place.
The first attempt
A skill
  • How to do it
  • A copy of the truth
  • A copy of the voice
The copies are the problem. Every copy drifts.
What teams move to
A skill
  • How to do it
reads
The context
  • The truth
  • The voice
One shared place, updated once. Every skill reads it.

That shared place holds your point of view, and every company needs one because the model has none of its own. Luckily, a lot of your point of view already sits in records you have. Sales calls are often the richest, because they hold your buyers' own words.

Your point of view, written down, has four parts.
IdentityYour voice, your values, and what you never claim.
AudienceHow each buyer thinks and decides.
FactsA verified record of what's true in your domain, which the model may not override.
MissionWhich assets go on which channels, and what each one is meant to achieve.
How Andrei does it

“I never write the point of view from a blank page. It gets mined from records the company already holds, and the client approves the facts before anything is drafted. For one digital insurance brokerage, that was about 400 reviews, roughly 1,000 sales call transcripts, and one executive workshop, and it came out as the four documents above.”

Andrei Matveev

The same files also record clearance, meaning which clients may be named and what stays confidential. So the model's limits are written down rather than assumed. And none of it gets written from scratch. Have AI draft it from the records, then have your senior people correct the draft. Correcting a draft is far less work than authoring one.

What changes

The content stops sounding like every other company and starts sounding like yours.

Split the context by how fast it changes

Some of the context changes every quarter and some of it changes every day, so it splits into two layers.

Two layers, split by how often the information changes.
The slow layer

Your point of view, written down in step 1.

Changes quarterly at most.

Kept in files that your senior people review on a schedule.

The fast layer

Deals, customer conversations, campaign numbers.

Changes daily.

Stays in your CRM and the other systems you already run. The AI reads it live through a standard connector, which the field calls MCP. The AI sees only what the person running it is allowed to see.

The slow layer only gets written if someone writes it, which is why writing it down was step 1. The fast layer already stays current inside your systems, so don't copy it into files. Connect the AI to those systems instead.

What changes

The model works from live numbers, and the systems you already run keep doing their job.

Tell the model how to read the context

A model handed a folder of documents decides on its own how to use them, and it decides differently every run.

So the harness should include a reading procedure, meaning a short document that tells the model how to read the other documents. Few guides mention this step, and it is the one that removes the guesswork.

The same folder of documents, read ten times.
Without a reading procedure
10
different readings of your strategy
With a reading procedure
1
reading, the same on every run
The reading procedure tells the model
  • Which document wins when they disagree
  • What order they're read in
  • What gets checked before a word is written
How Andrei does it

“Every harness I have built carries a reading procedure of a few pages, and it makes the AI state which document each decision came from before it writes a word. The brokerage's ran five pages.”

Andrei Matveev
What changes

Every run reads your strategy the same way, so the content stops changing from one run to the next.

Turn every rule into a check

The context files from step 1 say which clients may be named and which claims are off limits. Written as a sentence in a file, that rule depends on the model reading it and obeying it on every run. It usually does. For content that goes out under your firm's name, usually isn't good enough.

So every rule gets a check that is not the model. This is the step that stops claims nobody cleared from going out. There are three kinds of check.

Three kinds of check, from the strictest to the most human.
Code that blocksA hook is a small program that runs at a fixed point in the job and stops the model from doing what it was about to do. Nothing gets past it.
A separate checking passAnother AI run reads the draft against the rule. A single pass misses things, so the check runs as several separate passes. In practice this is usually a second agent, often on a different model, whose only job is to check.
A named personSomeone reviews and signs, and nothing ships without that signature.

The four rules below are examples from a working harness. Yours will be different, and each one still gets one of the three checks.

Four example rules, and the check each one gets.
Code that blocks

No client is named without clearance, and no off-limits claim goes out.

Blocks the draft before anyone sees it.
Code that blocks

Nobody already contacted gets contacted again.

Blocks the outreach batch before it starts.
A separate checking pass

Every fact is checked before it ships.

Sends a failed draft back before it reaches the person who signs.
A named person

A named person signs everything that leaves the building.

Nothing ships without the signature, and a record shows what shipped and what was claimed.
The second rule comes from outreach, which runs on the same harness.
How Andrei does it

“When the facts are approved first, expert review has little left to catch, and what it catches sits mostly outside the approved file. On the brokerage build, the client's experts and its agency's reviewers found eight corrections across eighteen web pages and twenty-one emails.”

Andrei Matveev

Every rule stays readable. Legal and leadership can open the rulebook, dispute any line, and see how it is enforced.

A lesson from Andrei's builds

“Readers miss things, and so do AI passes. On the brokerage build, two facts in one email contradicted the approved file. My checking pass and two rounds of human review all missed them, and an audit found them later. What would have caught them is code, meaning a check that compares every number in a draft against the facts file and blocks the draft on a mismatch.”

Andrei Matveev

Rules move toward the strictest check over time. A check that starts as a separate pass becomes code once it is clear enough to write down.

What changes

Drafts arrive close to final and safe to sign, so your senior people spend their review time on judgment calls instead of rewriting.

Make it the team's system

Up to here, all of this could live on one laptop. Once more people build their own skills, the copies multiply and some go stale. So the working rule is that everyone builds, and a skill gets shared only after review.

Everyone builds. A skill is shared only after review.
A personal skill

Anyone on the team builds one for their own job.

Review

Scored on quality, security, and overlap with what already exists.

The shared harness

Carries a named owner. The whole team works from the latest version.

What doesn't pass stays personal, and that's fine. A personal skill makes its owner better at their own job.

Once the harness is shared, the roles look like this.

Who does what once the harness is shared.
Named ownersEverything shared carries a named owner, so nothing is left to decay.
Non-technical peopleOpen a workspace that has been set up for them, with the latest version already in it, and work without touching the files and tools the builders work in.
A central teamOwns the volume jobs, an outreach batch say, and runs them as automated agents it owns. Personal skills stay personal, and people keep doing the parts that need their expertise.
What changes

Quality stops depending on who wrote the draft, and a new team member is productive on the latest version from day one.

Maintain it in releases

A harness goes out of date unless someone maintains it on a schedule. So it gets maintained the way software is, in scheduled releases, which content teams already know as sprints.

Three triggers, one release.
Time

A document hits its review date.

Decisions

A decision gets made and lands in the files instead of dying in a chat thread.

Drift

Live data contradicts a standing document. The audience file says one thing, the last twenty closed deals say another.

A release, a new version of the harness

The whole team works from it. Next cycle, the same three triggers say what to update.

A release ships only after it passes the same set of cases every time. The set is small, and the marketing team owns it. Engineers call this set an eval.

What every release is tested against.
Requests that must come out rightTen or fifteen real requests, each with an approved answer to compare against.
Requests that must be refusedA handful of requests the AI has to decline, such as a claim the facts file does not support.
Last cycle's correctionsEvery correction that repeated becomes a rule in the context and a new case here.
The set grows with every release, and a new model from the vendor has to pass it like any other change.

Keep the process light.

What changes

The harness stays accurate instead of going out of date, which is what makes it infrastructure rather than a one-off project.

Beyond one practice

One system, many practices and brands

A firm with several practices doesn't build a new harness for each one. It builds one core and derives a version for each practice from it.

One core, and a version for each practice or brand.
The core

House voice, group positioning, rules and checks.

Practice A

Its own audience, its own facts, its own missions.

Practice B

Starts from the core, so it is never built from scratch.

Brand C

An acquired brand keeps its own voice and still runs on the core's rules and checks.

Three versions, one core to maintain.

Data is never even across regions and practices, and that's normal. Start with the practice whose records are richest, and add the others after.

What changes

A change lands once in the core and reaches every brand, and the only differences between brands are the ones you chose.

The limits

What a harness can't do

Three limits worth stating plainly.
Can'tGuarantee results before publishing. Whether an asset gets results only shows afterwards, and that feeds back in through upkeep.
Can'tRun without upkeep. Facts go stale and models change, and a harness nobody maintains drifts back toward the generic content you started with.
Can'tReplace your judgment. It enforces the judgment you've stated and escalates what you haven't.
Where to start

One job, all the way through

Don't take all six steps at once. Pick one recurring job that eats real time, and build the whole path for that one job.

One job builds most of what every other job needs.
One recurring job

The one that eats real time.

Built for that job

Its context, its checks, its release schedule.

Reused by every other job

The point of view, the audience, and the cleared facts carry over.

After the first job, the shared folder looks like this, seven things in all. Each folder and file was built in the step marked beside it. The file the AI reads first sits at the top, and the model has no folder at all, since it is whatever vendor you connect.

The shared folder after the first job.
the harnessThe shared folder everything below lives in.
read firstThe reading procedure. Which document wins when they disagree, what order they are read in, and what gets checked before a word is written.Step 3
contextThe slow layer, meaning the four parts of the point of view plus clearance. Named owners keep it current, and corrections that repeat land here.Steps 1 and 6
identityYour voice, your values, and what you never claim.
audienceHow each buyer thinks and decides.
factsWhat is true in your domain, which the model may not override.
missionWhich assets go on which channels, and what each one is meant to achieve.
clearanceWhich clients may be named and what stays confidential.
connectionsWhere the AI reads the fast layer, meaning your CRM and the other systems you already run.Step 2
skillsOne folder per job, each holding the written instruction for how that job runs. A skill moves in here only after review.Step 5
checksThe hooks and the checking passes, one for each rule in the rulebook.Step 4
examplesThe cases every release is tested against. Requests that must pass, requests that must be refused, and last cycle's corrections.Step 6
recordWhat shipped, who signed it, and what was claimed where.Step 4

The models will keep getting swapped every few months. The harness stays through every swap, so the harness is what a team should spend its time building.

Sources

Where the facts come from