What an account-first AI sales agent does

The thelooma Sales Agent starts with a company account, not a list of email addresses. It helps a small sales team move from finding a promising company to understanding why the timing matters, choosing a sensible way in, managing the conversation and preparing a commercial offer.

The core flow is simple: find the right company, understand why now, choose the best way in, and turn the conversation into an offer.

One account holds the whole sales story

Most sales tools split the work into separate lists, inboxes, notes and documents. That makes it easy to lose the reason an opportunity mattered in the first place.

An account-first workspace keeps the opportunity verdict, research brief, people, contact-route evidence, message drafts, conversation history, meeting notes, next step, owner, proposal versions, offer versions and activity timeline together.

The action panel changes as the opportunity moves from research to contact, conversation, meeting and offer. Every open account should have one clear next step.

1. Find and filter promising companies

The agent can look across existing customers, past clients, unanswered enquiries, old proposals, referrals, event contacts, imported lists, inbound forms, booked calls, visible business signals and connected data providers. It can also look for companies similar to the seller’s best customers.

The point is not to create the biggest possible list. Weak matches are filtered before detailed research begins, so time goes to a small number of relevant opportunities.

2. Research the company with sources

For each selected company, the agent prepares a dated research brief. It covers what the company does, who it sells to, recent business signals, why the timing may matter, how it appears to handle the need today, likely objections and questions for the first conversation.

A possible business need is labelled as a hypothesis. Important claims carry sources. When reliable information is missing, the brief says that it is unknown instead of filling the gap with a guess.

3. Separate Fit, Timing and Way in

The agent does not compress an opportunity into a made-up win probability. It evaluates three different questions:

• Fit: How closely does the company match the ideal customer?
• Timing: Do recent signals suggest a reason to act now?
• Way in: Is there a suitable and supportable route to a relevant person?

Each area is Strong, Medium, Weak or Unknown and includes the reasons behind the rating.

4. Map the buying group and the available routes

The agent identifies relevant people and marks the likely role of each one: decision maker, influencer, user, gatekeeper or unknown. It uses relevant professional information only. When there is no reliable personal observation, the approach stays grounded in company research.

Possible routes include email, a warm introduction, a reply to an enquiry, a phone call, a manually sent LinkedIn message or a formal letter. For each route, the workspace records where the information came from, whether an address is generic or named, the recipient country, the available evidence and the current contact-policy verdict.

A discovered address is not treated as permission to send. Provider verification is not treated as permission either. The platform helps the user document evidence, understand the conditions that may apply and make an informed decision. It does not guarantee legal compliance.

5. Prepare one relevant approach

The agent prepares one editable message for one relevant person. The draft uses a verified observation, explains why it may matter to the seller’s offer, asks one small question and includes the required sender and opt-out information.

Checks look for unsupported claims, false familiarity, banned phrases, missing information and unnecessary length. Every first message waits for human review. LinkedIn remains manual: the platform prepares the draft and opens the profile, but the user sends it.

A refusal or unsubscribe request blocks future contact.

6. Support the conversation and meeting

Once contact begins, the agent can help interpret replies, prepare a response draft, suggest a stage change, prepare one appropriate follow-up and flag accounts without activity or a next step.

Before a meeting, it prepares a brief. After the meeting, it turns notes into confirmed needs, objections and promised actions. The user confirms important changes.

7. Turn confirmed needs into a proposal and offer

The Proposal and Offer Agent uses the completed research and confirmed conversation details to prepare a one-page recommendation, meeting recap, tailored proposal, scope, implementation plan, formal offer draft and the email used to present it.

It works from the company’s approved catalogue: products, packages, prices, included and excluded work, payment terms, delivery rules, approved proof, templates, brand voice and legal company details. It cannot invent a price, discount, delivery commitment, legal term, customer result or missing company fact.

A proposal explains the need, recommended solution, scope and plan. A formal offer contains the exact parties, scope, price, payment, delivery, validity and acceptance terms. The formal offer is prepared after the need and scope are confirmed, and a person reviews it before it is issued or sent.

What to do this week

• List the five companies you would most like to work with and write one sentence on why each fits.
• Add one recent signal that might make the timing relevant, or mark timing as unknown.
• Write down every possible way into each company, including warm introductions and past enquiries.
• Identify the person who reviews first contact and the person who approves price and scope.
• Collect your approved products, prices, payment terms and proposal templates in one place.

The Sales Agent page shows the full workspace. You can also see how it works alongside the social, ads, video and landing page agents. To map it to your sales process, book the free growth teardown. You will receive a clear next step, an exact quote and a delivery date.