Founders and sales

#The work around every call

A founder or a small sales team spends hours on the work around the actual conversations. They read up on a company before a call, write notes afterwards, check what competitors shipped, sort customer feedback and answer email. None of it is hard, but in a busy week it gets dropped.

This work suits BotsBot An AI teammate with a name, a job and its own conversation. It works on a computer in the cloud, even when you're away. See in the glossary → because it repeats, it reads from tools you already use, and the output can be a draft you approve. It also touches customers and other companies, so the boundaries from lesson 04 matter more here than anywhere else in the course.

#Six customer-facing Bots

Give each customer-facing job its own Bot. Let it prepare and draft, and keep every outgoing message behind your approval. The livestream session for founders showed six Bots built this way. They are demo Bots with invented customers, so treat them as patterns to copy. You won't find them as products to install. For official starting points, see the Sales Outbound, Account Health and Chief of Staff use cases in Use cases.

#Closebot: before and after the call

Before a call, the Bot pulls product usage for that customer account, researches the company and walks through its website on its computer. It returns a briefing as an HTML page (a web page you open in the browser). The briefing covers who they are, their licenses and recent meetings, a usage chart showing growth or decline, risks and recommendations. As a routineRoutine Work a Bot does on its own, on a schedule or when something happens in another tool. See in the glossary →, it reads the calendar each morning and prepares every meeting of the day.

After the call, it reads the meeting transcript (the demo used Granola, a meeting notes app). It notes what landed and what didn't, and adjusts future briefings. In the demo, this started paying off after two or three weeks.

#Prodbot: what shipped

A daily rundown of what was released and removed. The Bot reads pull requestsPull request A proposed change to the code, packaged so someone can review it before it's accepted. See in the glossary → and issues in Linear (a task tracker for software teams). Then it clicks through the product itself and attaches screenshots and a video of the walkthrough.

#Stockbot: competitor tracking

A "competitor pulse" routine that runs every few days. The Bot reads changelogs (the public lists of product changes), posts on X and job listings. It signs up for each competitor's product with a throwaway email, goes through onboarding (the first-use setup) and compares it with yours. It returns a teardown, a detailed breakdown of the product, with a video of the signup.

#Protobot: feedback to pull requests

The Bot pulls recent customer feedback, picks an item and launches a cloud agentCloud agent A Cursor coding agent that works on its own computer on a copy of the code and returns the change for review. See in the glossary → (lesson 15) to fix it. The goal is to gather every feedback channel into one Bot that opens pull requests, so a complaint becomes a reviewed change within hours.

#Yap bot: your voice

The Bot reads what you've sent by email and Slack to learn how you write, and relearns on a schedule. Other Bots ask it for help when they write on your behalf. Sensitive emails stay as drafts, and the Bot learns from the difference between its draft and what you actually sent.

#The catch-all Bot

You met this one in lesson 11. It takes one-off questions so they don't clutter the context of your specialist Bots. When it learns something useful, tell it to pass that on to the right Bot.

#Responsible use

Three rules from the docs apply to all of these Bots:

• Nothing goes out without approval. The docs' outreach prompts end with "do not send or enroll anyone" and "Do not contact customers". Add Ask first rules for external email under Settings → General → Auto-review.
• Respect the terms of the sites you use. The Sales Outbound and Talent Scout use cases say to use professional networks "as permitted by their terms" and to respect privacy and regional requirements. If a site blocks automation, the Bot should hand that step to you instead of working around it.
• Remember the shared computer. A voice Bot that reads your inbox uses sessions that every Bot on your account can use (lesson 02). Only sign in to what the job needs.

#Start with call prep

Start with the call briefing. It only reads data, and you see the value on the first day.

1. Create the Bot and write its job in the description: prepare briefings for customer calls, read-only, never contact customers or edit the CRMCRM The tool where a team tracks customers and sales deals. See in the glossary →. A CRM is the tool where your team tracks customers and deals.
2. Connect the sources. You need your calendar, the CRM or usage data, and the folder or tool where call transcripts land. Use Settings → Plugins where a connectorConnector A ready-made link between a Bot and a service such as Slack, Notion or GitHub. In the app they appear as Plugins. See in the glossary → exists.
3. Run one briefing by hand for a real upcoming call, and correct it until you would actually use it.
4. Save it as a skill (lesson 06) with the output format you settled on.
5. Make it a routine. Every weekday morning, the Bot prepares briefings for that day's external meetings.
6. Close the loop. After each call, ask the Bot to read the transcript and note what to change next time.
7. Add the next Bot only once this one is reliable.

#Closebot, Stockbot and Yap bot in the demo

Day 1 · 7:33:07Preparing a customer call with Closebot

In the demo, Closebot prepared a call with "Northwind", an invented customer. The briefing covered who they are, what their product does, screenshots of their website, account activity, a usage chart, risks and recommendations. The presenter added that the Bot often finds problems on the customer's site while it looks around, such as a cookie banner covering the submit button.

A version you can adapt:

Prepare a briefing for my 3 PM call with Northwind. Pull their product usage for the last 90 days, their open support tickets and my notes from past calls.
Research the company and review their website. Return a one-page HTML briefing with: who they are, account activity, a usage chart, risks, and three questions I should ask. Link every source. Do not contact anyone.

Day 1 · 7:46:25A competitor teardown from Stockbot

Stockbot's output for two invented competitors showed how a routine can end in one document instead of a stream of alerts. It held the signup video, what works and what doesn't, the changelog, posts on X, hiring and the team.

Day 1 · 7:50:30Yap bot writing in the presenter's voice

For a voice Bot, start narrower than the demo did:

Read the last 50 emails I sent from my work account. Write a short style guide of how I write: greeting, length, tone, phrases I use and avoid. Save it as a skill called "Write like me". Do not send anything.

#Messages that go out unreviewed

• Letting a Bot send outreach on day one. A wrong name or claim reaches a customer and can't be taken back. Fix: keep drafts and approval until you have reviewed many good ones. An approval doesn't reverse work already completed.
• Giving a voice Bot every account you have. Every other Bot on your account can now use those sessions. Fix: sign in only to what the job needs, and sign out when you stop.
• Automating competitor research without reading the terms. The Bot breaks a site's rules in your name. Fix: check the terms first and let the Bot hand blocked steps to you.
• Sending feedback straight into merged code. A single loud complaint changes the product. Fix: have the Bot open a pull request and review it like any other change (lesson 16).
• Asking random questions to specialist Bots. Their context fills with noise. Fix: use a catch-all Bot and pass on only what matters.

#Drafts always, sends with approval

• Give each customer-facing job its own Bot: briefings, shipped changes, competitors, feedback and your voice.
• Start with read-only call briefings, then add a routine and learning from transcripts.
• Every message to a customer or another company stays a draft behind Ask first rules.
• Check site terms and privacy rules before a Bot signs up for services or contacts people.
• Keep a catch-all Bot so your specialist Bots keep a clean context.