How AI decides who to name, and how we change it
Engines pick sources they can reach, understand, quote and verify. Each of those has specific work, and each can be measured.
Buyer asks
What is the best SOC 2 platform for a 60-person startup?
AI engines
AI answer
Waiting for sources…
The four layers engines read
Each layer depends on the one below it. Click through the stack.
Engines read from the bottom up. Click a layer.
Layer 1: Access
Can engines and agents reach and read you?
- AI crawler policy
- Server-rendered content
- Markdown for agents
- llms.txt and clean sitemaps
The first four months
A typical engagement. The audit can run on its own as the first two weeks.
- 1
Weeks 1 to 2
Measure
Prompt set, competitor baseline, citation source map and accuracy review. You see the starting line before anything changes.
- 2
Weeks 3 to 6
Fix the foundation
Access and facts first, because nothing else works if engines cannot read you. Then the pages closest to earning citations.
- 3
Months 2 to 3
Build and earn
New answer pages for prompts you are losing, plus outreach to the third-party sources the engines trust.
- 4
Month 4 onward
Compound
Weekly tracking and monthly priorities. We double down on what moves and cut what does not.
What we will do
- Show you raw prompt results, not just summaries
- Tell you when a tactic is not working and stop it
- Leave you owning every page, dataset and line of code
- Disclose any paid placement
What we will not do
- Guarantee what a model will say
- Buy links, write fake reviews or astroturf communities
- Publish AI-generated filler at volume
- Lock your data inside our tools
Find out what AI says about you this week.
Get a free AI Answer Snapshot: five of your buyer prompts, tested across six engines, with the competitors named instead of you.