Offer / ICP AI agent

I used to keep my ideal customer profile and offer notes in half a dozen docs. Every time I wrote an ad or trained ChatGPT I copied snippets across, tweaked pronouns, and prayed for consistency. Two years ago I replaced the chaos with a single living one-pager linked to a custom GPT. The result: faster copy, aligned pitches, and zero mixed messages to the market.

This chapter walks you through building the same system. First we draft a concise ICP & offer sheet. Then we park it in Notion where the whole team can update it. Next we craft a prompt that feeds any chat model the right context. Finally we wrap the prompt into an AI agent so fresh, on-brand copy is one slash-command away.

Create a one-pager that covers four essentials: who the buyer is, which job they hire you to do, the core outcome you guarantee, and the proof that de-risks the purchase. Strip fluffy adjectives. Write in bullet points so models parse it cleanly.

Example layout: buyer profile → growth-stage SaaS CMO with €5–25 million ARR. Job-to-be-done → increase qualified demo volume without expanding team. Core offer → ninety-day demand engine built around LinkedIn and paid search. Proof → case study links and quantified lifts.

Keep the page under 300 words. Large context blocks slow language models and invite hallucination. Precision beats paragraphs.

Bridge: with the one-pager tight we need a shareable home, so next we store it in Notion.

Create a new Notion database called “Offer & ICP Library.” Add columns for Version, Last Edited, and Owner to maintain hygiene. Paste the one-pager into the page body, using heading blocks for each section so anchors stay stable when content grows.

Turn on page history so changes and rollbacks are painless. Mention the doc in your onboarding checklist for new hires; marketing, sales, and delivery should all know where the truth lives.

Copy the public page link—this will feed everything from chat prompts to proposal templates. If you prefer ClickUp or Confluence, the flow is the same; only the URL changes.

Bridge: the source is set, now we need a prompt that fetches and formats it on command.

Open your favourite chat model and paste the following skeleton prompt:

“You are my growth copy assistant. Use the ICP & Offer at [Notion URL] as the only source of truth. Whenever I ask for copy, first summarise the buyer’s pain in two lines, then write the requested asset. Never invent new features or audiences.”

Tweak tone, length, and formatting rules to suit house style. Test with a quick request—“Draft a LinkedIn hook” or “Write an email subject.” Review the output. If it drifts, tighten the prompt or trim the one-pager until responses snap back to reality.

Bridge: a static prompt is handy, but an AI agent saves even more time, so we automate it next.

Many tools let you package prompts into reusable commands. In ChatGPT create a new GPT, paste the refined prompt, and set the Notion link as a permanent instruction. In Zapier you can build a workflow that pulls the page text via API and feeds it to OpenAI whenever someone types “/offer email.”

Add role options—copywriter, SDR, or strategist—so the agent tailors language without breaking alignment. Finish by sharing the GPT or Zap with the team. One click now delivers headlines, outreach snippets, or proposal intros that all sing from the same hymn sheet.

Bridge: the agent will only stay sharp if the source stays current, so the conclusion covers your maintenance rhythm.

A single, up-to-date Offer & ICP one-pager fuels every touchpoint when it is stored centrally, wired into a clear prompt, and wrapped inside an AI agent. Build the page today, link it, and automate the fetch. Tomorrow your ads, emails, and chat responses will share one voice and one promise—freeing you to refine strategy instead of chasing consistency.