Sales & CRM

Sales with AI agents: capture inquiries, draft quotes, follow up.

Most sales time goes into work that has nothing to do with the conversation: capturing inquiries, putting quotes together, updating the lead after the meeting, remembering to follow up. An agent does this by your rules and asks you before anything goes out. Here is a week with four such runs.

A week in sales

Four agent runs in one week. Each starts from a trigger, connects several systems and ends with a result, not with one more manual step.

  1. 01 · Mon 08:20

    Inquiry, lead and quote

    At 08:20 the inquiry from Lindworks arrives at sales@, with the requirements list attached. The agent puts the lead on the existing company, confirms receipt and drafts the quote from the price list and the framework contract: 240 × B-40, €17,556.00. After Sarah's yes in Slack, QUO-2026-0417.pdf goes to Petra Keller.

    1. 1The inquiry arrives at sales@, with a requirements list
    2. 2Lead created, the existing company recognised
    3. 3The acknowledgement goes out in the same minute
    4. 4Quote drafted from price list and framework contract
    5. 5One yes in Slack, then the PDF goes to the customer
    InboxRecordPrice listSlackEmail to customer
  2. 02 · Wed 07:30

    Briefing before the customer meeting

    The meeting at Lindworks is at 10:00, and the schedule sees the date on the lead. The agent pulls the open promises from the lead, the last emails from the inbox and past orders from the records and files a one-page briefing as a page in the space. At 07:30 it is in Sarah's Slack, before she drives off.

    1. 1The meeting date sits on the lead
    2. 2Lead, last emails and orders pulled together
    3. 3Open promises from the last meeting go on top
    4. 4The briefing is in Slack at 07:30
    ScheduleRecordsInboxPageSlack
  3. 03 · Wed 17:40

    Three lines in Slack keep the lead current

    After the meeting Sarah sends three lines in Slack. The agent puts them on the lead: stage to "quote negotiated", next step sample B-40 by 25 Sep, note and stage also go to Attio. If the sample has not gone out by 23 Sep, it reminds her.

    1. 1Sarah sends three lines in Slack after the meeting
    2. 2Stage and next step sit on the lead
    3. 3Note and stage go to Attio
    4. 4The reminder for the sample is set
    SlackRecordAttioReminder
  4. 04 · Fri 08:00

    Following up on everything still open

    On Fridays one run checks every open quote and every lead. For seven cases with no answer for seven days the agent drafts a follow-up email with the context from the inquiry, and Sarah releases all of them with one yes. Whoever stays silent in the second round goes cold, with a reminder in 60 days.

    1. 1One run checks every open quote and every lead
    2. 2No answer for 7 days: a draft with context
    3. 3Sarah releases the seven emails with one yes
    4. 4Emails out, records and Attio updated
    5. 5Silent twice: cold, with a reminder in 60 days
    ScheduleRecordsSlackInboxAttio

Common questions

No. Records is the working database for inquiries, quotes and meetings, where the agent and the team see the same state. Attio is connected as an integration, another CRM through MCP or an HTTP step. Small teams use Records only, larger ones have stage, note and next step written into the CRM.

From your own documents: the price list as a document or Google Sheet in the space, the terms from the customer record. The skill calculates with those instead of inventing prices. If it finds no price for a line item, it flags that line and asks before the quote goes out.

Quotes and follow-ups only go out after a yes in Slack, and the draft is ready before that. For one narrow category, such as the acknowledgement without prices, you can allow sending so an inquiry does not sit unanswered until the afternoon.

Uncertain fields are flagged instead of accepted, and if something is missing the run stops and asks in Slack. Every run is logged step by step: you see which email, which document and which price it used. Changes to records can be undone.

Describe the task. The rest takes shape on the platform.

Start for free, create your first skill with the assistant, and see what a run looks like before you connect anything.