The problem this solves
Intent data is most useful in the first hours after it moves, and least likely to be looked at then. The dashboard is on a laptop; the seller is on a train. By the time the feed is checked, the five companies that surged overnight are mixed in with two hundred that did not, and the morning is spent sorting rather than reaching out.
A digest solves the reading problem by doing the sorting first. Five companies, chosen by the size of their overnight change on <topic>, one line each explaining who they are and why they surfaced, and a link to the full list for anyone who wants more. Delivered to Telegram at a fixed time, it is read on the phone before the first meeting, and the first touch of the day goes to the account that moved most.
The discipline is in what the digest leaves out. Flat scores, familiar names and companies outside the ICP are excluded, so the five are always new information rather than a repeat of yesterday.
How the mission runs
- Read overnight intent on the topic. Every morning the Signals Agent queries intent data for activity on <topic> since the previous digest, capturing the score per company and the change against each company's prior level.
- Rank by new intent and filter. Companies are ranked by the size of the overnight change rather than the absolute score, so the digest reflects what moved. Firmographics remove companies outside the ICP, and companies named in the last seven digests are excluded so the five are new.
- Write one line of context per company. For each of the five, the agent writes a single line: what the company does, its size, and the intent change, drawn from firmographic data and the intent topics. Where a related signal exists - a recent round, a relevant posting - it is folded into the line.
- Send to Telegram with a link to the full list. The five lines are sent to your Telegram bot at 8am in your timezone with a link to the full ranked list in the workspace, and the day's list is saved so the exclusion window works the next morning.
The prompt
This is the exact objective the agent receives. Swap the obvious placeholders for your own domain, segment or channel and run it as-is from the console, Slack, or the API.
What comes back
A Telegram message every morning at 8am with the five ICP-fit companies whose intent on <topic> rose most overnight, each with a one-line description and the intent change, plus a link to the full ranked list. The full list is saved daily with the scores and changes, so a week of digests can be reviewed as one table.
Make it yours
- Ask for the likely buyer with a verified email in each line, which turns the digest into a calling list.
- Send the same digest to a Slack channel or a WhatsApp number instead.
- Track two topics and get five companies for each, labelled.
- Change the time to match your first block of outreach, or send twice a day during a campaign.
Frequently asked questions
What if fewer than five companies moved overnight?
The digest sends the ones that did and says so; it never pads the list with flat scores to reach five.
Can I reply to the digest to get more on a company?
Yes. Reply in Telegram with the company name and the agent answers in the same chat with the full profile, the buyer and a suggested opener, and can save the profile to Notion.
How is the ICP applied?
From your saved ICP or the criteria in the prompt; the digest names the filter it applied so you can loosen or tighten it.