Strawberry Lead Research FAQ: Sources, Sheets, Approvals

The example is made up. Dana Whitfield runs sales at Northpoint Payroll Partners, a 9-person firm that sells outsourced payroll to architecture and engineering firms. Every week her team loses hours researching firms into a sheet and writing first emails by hand. Every firm, person, address and number in the lesson is invented.

A note on the screens: the lesson was built from Strawberry's docs and website, read on 21 October 2026, without installing it. Strawberry 1.0 launches on 26 October 2026, so labels can differ by version. This is our own lesson, not an official Strawberry course.

How does the Strawberry lead research workflow work?

The lesson follows five parts:

1. Set the guardrail. Type strawberry://settings/integrations in the address bar, open Gmail, and under Permissions change Send messages from Always allow to Ask every time.
2. Research the leads. Click New Tab and paste a prompt asking for 10 Ohio architecture and engineering firms with 20 to 80 staff that are hiring an office manager or HR person, with an evidence URL and a confidence for each.
3. Check the work trace. Open the row of actions to see every search and page the companion read.
4. Save to a sheet. Drop the firm that is already a client and save the other 9 rows to a new Google Sheet called Q4 outbound - A&E firms, with a Status column set to New.
5. Draft, check, approve. Ask for three first emails, fix a wrong fact on the approval card, approve one send, then turn the chat into a skill.

Pulling company details from public sources into a list is a form of lead enrichment, and knowing who you sell to first is your ideal customer profile.

What makes a good research prompt?

Strawberry's research guide says to start with the decision the research supports, say which sources to use first, ask for a table with an evidence URL and a confidence on every row, and accept "not found" instead of a guess. Ask for 10 rows, check them, and only then ask for more. The lesson's prompt also ends with "Ask me before sending any outbound messages."

The result in the lesson shows why. Ten firms come back, plus a short list of gaps: one firm's staff size was not found, and one fact came from a local blog, not the firm itself, so it is marked low confidence.

Key takeaways

• Check the connected app's permissions before any agent work. Sends go on Ask every time.
• Ask for sources and confidence, accept "not found", and start with 10 rows.
• Open the work trace and check a few rows against their sources.
• Read every email card against the sheet, fix it, and only approve what you would sign.
• Save the workflow as a skill, and make sure the approval rule is in it.

Frequently asked questions

What is a Strawberry companion?

An AI helper inside the Strawberry browser. It can search, read pages, use your connected apps and take multi-step actions, using your own tabs and logins. That is what makes it an AI agent rather than a chatbot.

Where do I set what the companion can do in Gmail?

At strawberry://settings/integrations. Open the app's page and set each listed action to Ask every time, Always allow or Disable action. Saved permissions apply without a new prompt, so check them first. The docs don't name each Gmail action, so the lesson uses plain wording such as Send messages.

Which actions should stay on Ask every time?

Strawberry's docs suggest starting read-only: let it read and draft, and keep sends, record changes, deletions and anything that spends money on Ask every time. These are agent permissions, and they are the main guardrail.

What is the work trace?

The row of actions under the companion's answer. Open it to see each search, page and app action it took. In the lesson, it shows official sites first and one blog it marked low confidence. More in the work trace entry.

Why ask for a confidence on every row?

A list of firms with no sources is a guess. A row with an evidence URL is something you can check in ten seconds, and a confidence tells you which rows to check first.

What does the approval card look like?

When an action is on Ask every time, Strawberry shows a request above the chat input. Email cards are editable: recipient, subject and body. In the lesson, the card says a 40-person team but the careers page says 24, so you fix the body before you click the check mark. The docs describe the card but not its button labels, so the lesson keeps that screen plain.

Can I take back an email after I approve it?

No. A sent email cannot be taken back. Before you approve, check the account, the recipient and the facts against the source.

What should the first emails say?

The lesson asks for under 90 words, one fact from the firm's evidence URL, a plain voice and no claims about price or savings. Keep a cold email short and checkable.

How do I repeat this next week?

Ask the companion to create a skill from the chat, including the questions it asked, the sources it checked, the approval rules and the output format. Read the skill before you keep it: the last rule should be "Never send. Show each email for approval first." Only turn it into a routine once you trust the output.

Which account will the companion use?

The one connected in Strawberry. Built-in app connections are shared across your Strawberry profiles, so check which account a task will use before it touches client data.

Is Strawberry free?

Its site lists a free plan with monthly AI credits and paid plans, and normal browsing uses no credits. Check strawberrybrowser.com for current plans. The desktop browser runs on macOS and Windows.

What goes wrong most often?

Leaving sends on Always allow and approving an email without checking its facts. See Strawberry lead research mistakes, then try the practice.