Capability
GTM kernel
The machine-readable GTM record every agent reads before it acts.
One kernel, four agent families. Every agent reads the kernel before it acts.
- ICP tier
- [segment · firmographic]
- Positioning claim
- [the claim]
- Proof point
- [evidence + source]
- Competitive claim
- [vs. alternative class]
What is a GTM kernel?
A GTM kernel is a machine-readable record of a company's go-to-market (ICP, segments, positioning, offers, proof points, competitive claims, objection answers) that AI systems read before acting. [Product]'s kernel adds [differentiator], so every agent shares one governed source instead of ten conflicting ones.
How does the kernel work?
The lifecycle that makes “source of truth” credible: how the record is built, who governs changes, and which systems read it.
- 01
Ingestion
[How the record gets built, and from what sources.]
- 02
Versioning & approval
[Who governs changes: the story that makes 'source of truth' credible.]
- 03
Consumption
[Which agents + external surfaces read it: APIs, MCP.]
What does the kernel actually hold?
- ICP tier
- [segment · firmographic]
- Positioning claim
- [the claim]
- Proof point
- [evidence + source]
- Competitive claim
- [vs. alternative class]
- Objection answer
- [the response]
- Brand context
- [a governed subset, never the headline]
Before
- Generic
- [bland, off-positioning output]
After
- Kernel-grounded
- [on-ICP, on-claim output]
What jobs does the kernel power?
Every other capability grounds on this record before it acts.
Who controls what the kernel says?
Frequently asked questions
Who can change what's in the kernel?
How do agents read it, API or MCP?
Does the kernel train your models?
See the machine on your own GTM
[What happens next, one line under each path: the mechanism you just read, run on your data.]