A GTM playbook as code is your go-to-market written as plain files that AI agents read and follow: the ideal customer, the objections and answers, the method for each campaign type, the rules, and where a person approves. Like software, it is versioned. Unlike a sales playbook document, it is run every time, so it stays accurate.
Key takeaways
- A GTM playbook as code turns the go-to-market method into files agents follow, not a document people skim.
- It holds four things: who you sell to, what you say, how each job is done, and the rules and gates.
- Methods can be packaged as skills: according to the open Agent Skills standard, a skill is a folder with a SKILL.md file of instructions.
- According to Panopto's 2018 survey, 42% of institutional knowledge is unique to the individual who holds it, which is the knowledge a playbook captures.
- Versioning means every change is recorded and reversible, and the playbook improves from results.
What does GTM playbooks as code mean?
It means writing your go-to-market down in a form that machines can run and humans can review. Each part of the method, such as how an outbound campaign is built or how a follow-up is written, becomes a file of instructions. Agents read those files before doing the work, and changes are tracked like code.
The phrase borrows from software engineering, where infrastructure and processes are written as files so they can be versioned, reviewed and reproduced. Applied to go-to-market, it means the method is no longer in people's heads or in a slide deck. It is in files the business owns.
The practical benefit is consistency. Every campaign is built the same way, every follow-up follows the same rules, and when someone improves the method, everyone's next campaign uses the improvement.
What goes in a GTM playbook?
A GTM playbook holds four kinds of file: who you sell to, what you say, how each job is done, and the rules and approval points. Together they let an agent run a campaign the way your best people would, and stop where a person should decide.
| File type | Holds | Example |
|---|---|---|
| Who | Ideal customer profile, tiers, disqualifiers, signals | Tier 1: multi-site services firms of 50 to 300 |
| What | Objections and answers, customer quotes, approved claims, voice | Proven answer to "we use spreadsheets" |
| How | Method for each job, as skills | How a first-touch email is written |
| Rules | Volumes, approvals, what AI may never do | Campaigns load paused; a person launches |
The "how" files are where skills fit. According to the open Agent Skills specification, a skill is a folder containing a SKILL.md file with a name, a description and instructions, which agents load when a task matches.
How is it different from a sales playbook document?
A sales playbook document is written for people to read, usually once, and drifts out of date. A playbook as code is written for agents to follow on every job, so wrong instructions show up in the output and get fixed. It is also versioned, so changes are visible and reversible.
Most sales playbooks share a fate. They are written during a strategy push, presented, filed and forgotten. Reps ask colleagues instead. Within a year the document describes a process nobody follows.
A playbook that runs every day cannot drift quietly. The guide to process documents versus skills explains why use keeps written methods accurate.
How do agents use a GTM playbook?
Before doing a task, an agent reads the relevant files: the ideal customer to decide fit, the objections and quotes to decide what to say, the method to decide how, and the rules to decide what it may do and where to stop. It then does the work and hands the result to a person.
Anthropic's guidance on building agents favours simplicity: find "the simplest solution possible, and only increasing complexity when needed". A playbook as code fits that advice. Most GTM work becomes predictable workflows with an AI step, guided by clear files, rather than free-roaming agents.
The knowledge in those files is what makes the output good. According to Panopto's 2018 Workplace Knowledge and Productivity Report, 42% of institutional knowledge is unique to the individual who holds it. A playbook as code is how that individual knowledge becomes the company's.
How do you start writing a GTM playbook as code?
Start with one campaign type you run often, such as outbound to your top tier. Write the ideal customer file, the objection file and the method for that campaign, plus the rules. Have an agent run a campaign from them, review the output, fix the files, and repeat. Expand to other campaign types once it works.
A first month:
- Week one. Write the ideal customer profile and disqualifiers.
- Week two. Write the objection library and approved claims.
- Week three. Write the method for one campaign as a skill, with an example.
- Week four. Run it, review every output, and fix the files.
Version control matters from the start. Keep the files in a shared, versioned folder so every change is recorded with who made it and why.
Skein builds GTM playbooks as code inside a Company OS, with the method as skills and the rules enforced at the approval points; the AI engineering page shows a campaign running from them. That suits a B2B team that wants its go-to-market to survive staff changes. A founder still finding product-market fit may want to keep the method loose for now.
Frequently asked questions
What is a GTM playbook?
A GTM playbook describes how a company finds, reaches and wins customers: who it targets, what it says, how each campaign is run and the rules it follows. Written as code, it becomes files that AI agents read and follow on every job, kept current and versioned.
What is the difference between a sales playbook and GTM playbooks as code?
A sales playbook is usually a document written for people to read, which drifts out of date. GTM playbooks as code are files agents follow on every task, so errors show up in the output and get corrected. They are also versioned, so changes are tracked and reversible.
What should a GTM playbook include?
It should include the ideal customer profile with tiers, disqualifiers and signals; objections with proven answers, customer quotes, approved claims and voice; the method for each campaign type; and the rules, such as volumes, approvals and what AI may never do.
Can AI run a go-to-market playbook?
AI can run much of it: researching accounts, drafting outreach, building campaigns and reporting, all guided by the playbook's files. A person should still approve and launch campaigns, handle replies and decide on anything involving price, discounts or contractual commitments.
How do you keep a GTM playbook up to date?
Make agents use it on every job, so wrong instructions show up quickly in the output. Keep the files in a versioned folder, name an owner for each, and feed results back, such as winning messages and new objections, after each campaign.
Related reading
- Agent Skills. "Agent Skills Overview." agentskills.io, accessed 30 October 2026. https://agentskills.io
- Anthropic. "Building effective agents." Anthropic Engineering, 19 December 2024. https://www.anthropic.com/engineering/building-effective-agents
- Panopto. "Workplace Knowledge and Productivity Report." 2018, reported by HR Dive. https://www.hrdive.com/news/inefficient-knowledge-sharing-costs-large-us-businesses-47m-a-year/527892/