Guide · working with your AI
areyoufoundbyai.com/monitor/a1b2c3d4e5f60718The button in your weekly email opens a new conversation with that week's numbers pasted in as text. Useful once, but it means every week starts from zero, no memory of what you and your AI already worked through. There's a better way, and it costs you nothing to set up: your monitoring link is a live URL, not a snapshot. Fetched today it returns today's data; fetched again in six weeks it returns six-weeks-from-now's data, same link. Point your AI at that link once, as a standing instruction, and every new chat starts current automatically.
The launch film shows the loop this guide sets up: the engines answer your buyers, the weekly measurement records what they say, and your own AI works from that data once it is connected.
The film runs seventy seconds and has sound. If the player does not load, open the film directly (MP4, 12 MB).
Most AI chat tools let you set persistent instructions for a project or space, text applied to every new conversation inside it, not just the current one. Paste this:
Exact menu labels move around as these products update, look for "instructions," "custom instructions," or "project knowledge" inside whatever the tool calls a Project/Space/Gem. The paste-in text above is the part that matters and won't change.
The standing instruction above makes your AI read your data at the start of a chat. Your console is also a live MCP server, the open standard AI tools use to call other software, which means your assistant can query your measurements mid-conversation, whenever the question comes up, not just at the start. Your endpoint:
claude mcp add --transport http foundbyai-demo https://areyoufoundbyai.com/mcp/demoIt exposes sixteen read tools and one action, each a live query against the same measurements your console renders:
get_visibility, your current AI Visibility and AI Readiness scores (each /100) with the week before for comparison, your separate off-site Footprint score, and the subscores behind them.get_question_trajectories, every tracked buyer question with its measured history, day by day: how many of the 7 engines named you each time it was asked.get_rivals, the competitor names the engines actually gave buyers in the latest scan when they didn't name you.get_share_of_voice, your slice of AI-answer mentions against those rivals, and how it has moved over the tracked window.get_mentions, the new pages on the web that mention your business, from the daily sweep (pass days to set the look-back window).get_context, the whole weekly picture in one call: scores, questions, mentions, fixes, the same content as the context pack above.get_fix_plan, the prioritised fix list from your latest deep scan, highest leverage first.get_citation_sources, the domains the engines actually cite in your category, and whether you appear on each.get_personas, the buyer personas behind your tracked questions, with the exact questions each persona asks.get_post_brief, a drafting brief assembled from this week's verified data: fresh mentions, the questions you're absent from, the sources the engines read. Your AI writes the post, in your voice.get_ai_traffic, the visitors AI engines actually sent to your site in the last 30 days, recorded by your site's own beacon.get_agent_view, how a headless agent browser renders your site: full, partial or blank, with the characters an agent can extract.get_benchmark, where you rank against every other measured business in your category.request_rescan, the one action: queues a fresh measurement right now (capped by your plan's on-demand allowance) and changes nothing else.claude mcp add --transport http foundbyai https://areyoufoundbyai.com/mcp/YOUR-TOKEN Anything you'd ask a colleague who had your console open. The point of the live connection is that these work mid-conversation, weeks after setup, and always answer from this week's data:
If a coding agent builds your site, the next step is Connect your AI and let it work the plan: your agent reads the fix plan over this same connection, applies the fixes in your repository with your approval, and re-measures until a target you set.
Sixteen of the seventeen tools only read. The one action, request_rescan, re-runs your own measurement inside your plan's cap and touches nothing else: the connection can never change your account, plan or billing. The same privacy rule below applies, the URL contains your token.
Both formats carry the same live picture: your AI-visibility and readiness scores, the specific buyer questions engines do and don't name you for (with movement over time once you have history), the scanner's own ranked list of what to fix first, a citation game plan of where to get listed or reviewed so AI cites you, new independent mentions found in the last week, and adjacent questions worth tracking. The plain-text version at /context.md is built for attaching to a chat; the .json version is for anything that wants to parse it directly.
Keep the weekly email, the new-chat button is genuinely useful the first time, or any time you want a clean-slate read. The connected project is for the ongoing conversation in between.