Go-to-market when both sides use AI.
- 1 of 8: Discovery
The buyer asks, an assistant, not Google
A need, a team size and a budget in one sentence. The assistant decides where to look.
51% of B2B software buyers start research in an AI assistant. G2, 1,076 buyers, Mar 2026 - 2 of 8: Discovery
An agent reads, your real pages
Pricing, docs, reviews and forums. What it cannot read, it cannot quote.
Read: Your next buyer is an agent97% of llms.txt files got zero requests in a month. Ahrefs, 137,210 domains, Jun 2026 - 3 of 8: Discovery
Two or three names, come back
The vendors it could describe with confidence. Everyone else is invisible.
69% picked a different vendor than planned because of what an AI assistant told them. G2, 1,076 buyers, Mar 2026 - 4 of 8: Timing
You see when, not who
AI agents fetching your pages and public events replace content clicks as the timing signal.
Read: When research goes dark3 major assistants mark fetches made for a user's question. OpenAI, Anthropic, Perplexity, bot documentation - 5 of 8: Outreach
A model reads, your email first
Inboxes summarize, rank and archive mail before a person opens it.
Read: A model reads your email first73% of B2B buyers actively avoid suppliers who send irrelevant outreach. Gartner, 632 buyers, Jun 2025 - 6 of 8: Outcome
A first call, accepted
The step where both sides still talk in person.
- 7 of 8: Thesis
Agent meets agent, before anyone calls
Our read for 2027: the buyer's agent compares offers and asks the vendor's agent questions before a call is booked. Not measured yet.
- 8 of 8: Measurement
Which step won the call?, the last click sees the end
Most of the purchase happens in chats and on pages you never see. Analytics credits the last click, and visits from an assistant often arrive with no source and land in Direct.
61% of the buying journey is behind the buyer before they first contact a seller. 6sense, nearly 4,000 buyers, Nov 2025
Notes
- Discovery: How AI assistants build a vendor shortlist: What they read, in what order, and what makes them name you.
- Timing: Signals that survive the dark funnel: Which events still tell you when to reach out.
- Outreach: Write the email the summary keeps: If the point does not survive one AI sentence, it does not reach the buyer.
- Discovery: llms.txt: what the server logs actually show: The file most guides recommend, and the data on who reads it.
- Discovery: Write your pricing page for an agent: Ranges, scope and limits a machine can quote back.
- Timing: Reading AI user agents in your server logs: Which fetches mean someone asked about you, and which are crawlers.
- Outreach: Bulk sender rules on one page: Gmail, Yahoo and Microsoft: what gets mail rejected outright.