AI Agent for Lead Research: One Trello Card per Qualified Lead

A small company rarely lacks leads to look at. It lacks the hours to look at them properly. Studying an account, finding the right role, understanding the problem you could solve and deciding whether it is worth a call: each step is worth doing and none is urgent, so the list of target accounts slips behind client work week after week. This is the kind of background research an autonomous AI agent can carry, because it asks for patience and method rather than judgment.

The inspiration: a three-person company and its agent

The starting point is a public account from Halfbit Studio, a software house of three people, which describes how it uses an agent as a working partner. Among the tasks it hands over is the study of prospects. According to that account, the agent examined 24 leads and produced, for each one, a sheet with a temperature of hot, warm or cold, then one ticket per lead in the team's tracker.

What matters is not the volume but the shape of the result. Each lead becomes an object the team already knows how to handle, a ticket, instead of a line in a spreadsheet nobody reopens. The same account is also clear about the limit: the agent does not write to prospects, and it does not decide what to do with what it found. Atako applies the same split.

What the agent does

You give the agent the target accounts you listed, up to ten. For each one, it builds a sheet:

• what the company does, its size and the technology visible from the outside;
• one concrete problem your offer could address, one rather than a list;
• the role of the probable decision maker, a job title and not a person it cannot verify;
• an angle for a first contact: what to say and why now;
• a temperature, hot, warm or cold, justified in one line, with its sources as links.

It then creates one Trello card per lead in the list "Leads to qualify" of the board you named, with the sheet in the description and the temperature in the title, for example "[Hot] Acme Foods". It checks the existing cards first, so a company is never filed twice. When information is missing or uncertain, the card says so instead of guessing.

Every Monday at 9:00, the agent looks at the cards that have not moved for four days and adds a comment with a concrete follow-up suggestion.

What the agent does not do

The limits are part of the design. The agent does not send emails or messages to prospects, it does not book meetings, and it does not decide which leads deserve a call. It does not touch your CRM. Its only write access is the Trello board you chose and the Atako project created for the scenario. A hot temperature is an opinion based on public signals, not a verdict: the person who knows the market and the relationship stays in charge of the next move, which is the principle behind human-in-the-loop design.

How to set it up in Atako

From the page of this use case, the deployment takes four steps.

1. Connect Trello. Create an API key and a token from your Trello account and paste them when Atako asks. The connection is checked before anything is created. If you are not an administrator, the connection stays personal to you.
2. Fill in your context. Describe your offer in a few sentences, list the target accounts, one per line, and name the Trello board that will receive the cards. These inputs are what the agent works from, so the more concrete they are, the better the sheets.
3. Deploy. One action creates the agent and the "Lead research" project. The agent is provisioned and receives its first instructions on its own.
4. First results, then the weekly rhythm. The first cards appear as the agent studies the accounts, and it summarizes the leads in the chat with their temperature and a link to each card. From then on, it plans its own Monday follow-up.

If you do not have an account yet, creating it starts your trial and the scenario is configured right after. If your administrator has not allowed you to create agents, the deployment tells you so and stops there.

Where to see the results

You can follow the work in three places. In Trello, each lead is a card with its sheet and temperature, and you sort or filter the board as you would any other. In the Atako project "Lead research", you find the short how-to written for your team. On the Atako floor, the agent is visible with its current activity, so you know whether a run is in progress without asking.

You can also talk to the agent in the chat to ask for a new account to be studied or a sheet to be reworked. The request is handled in the same conversation, with the same memory of what was already done.

Safeguards

Access is narrow on purpose. The Trello connection is granted to the agent with a limited list of eight tools, enough to read the board and create or update cards. An agent that sees a short list of tools uses them more reliably than one lost in a catalogue of hundreds. Nothing is granted by default, every grant can be revoked, and the agent has no access to your mailbox or to your CRM.

Human validation sits at the decision points. You read the cards before anyone contacts a prospect, and you adjust the temperature when your knowledge of the market contradicts the agent's. The agent prepares the ground; it does not act on it.

What to expect

Expect a first pass that gives you a structured view of your target accounts, not a finished sales plan. The value comes from the second and third weeks: the board grows as you add accounts, stalled leads get a nudge without anyone remembering them, and the sheets get sharper as you correct them in the chat. For a team of two or three people, that is the difference between prospect research that is always on the to-do list and research that simply exists when you need it.