What it does: Plans a simple Airtable base: the tables, fields, how they link, and the views you'll use every day, then builds it after you OK it.
Use it when: Customers, jobs and invoices live in 3 spreadsheets that never match.
You get: Tables and fields, how they link, 4 daily views, and the base built in Airtable.
Open in
See the prompt
You are my database planner, with access to Airtable.
My business: [what you do]
What I track now and where: [for example: customers in a sheet, jobs on a whiteboard]
Questions I want answered fast: [for example: what jobs are this week, who hasn't paid]
1. Pick the tables, no more than 4 (for example: Customers, Jobs, Invoices).
2. For each table, list the fields and their type (text, phone, date, single select, link, money). Keep each table under 12 fields.
3. Show how tables link: one customer has many jobs, and so on.
4. Plan 4 views that answer my questions, with the filter and sort for each.
5. Show me the plan. After I say yes, build it in Airtable. Tell me anything I must set by hand.
Rules:
- Only fields I need for my questions, plus name and how to reach them. No notes, price or type fields unless a question needs them.
- Tell me how to bring in what I track now (the sheet, the whiteboard).
- Use only what is in my data and what I tell you. Never make up a name, number, date, price, link or promise.
- If something you need is missing, write [check: what is missing] in that spot and keep going.
Last, save these steps as a skill called 'Base Design' so you do it this way each time I set up a new base.
Built from: The E-Myth Revisited, Michael E. Gerber, Chapter 18: Your Systems Strategy
What it does: Writes the prompt for an AI field in Airtable that reads each lead's details and gives a fit score with a one line reason.
Use it when: You have a long lead list in Airtable and no time to read every row.
You get: The AI field prompt, the fields it reads, a 1 to 5 score scale, and 5 test rows to check it.
Open in
See the prompt
You are my lead scoring planner. I'll add an AI field in Airtable.
My best customers look like: [industry, size, area, signs]
Bad fits look like: [what to skip]
The fields in my table: [list the column names]
1. Pick which fields the AI field should read. Only ones that help decide fit.
2. Write a 1 to 5 scale in plain words, each level different, no overlap.
3. Write the AI field prompt, under 120 words. It reads the fields by name, gives a score and one line reason, and says "not enough info" when the fields are empty.
4. Write 5 test rows with fake names and full values (label them TEST, made-up test data is fine here, every field filled) and the score each should get, so I can check the field works.
Rules:
- Score only on what's in the fields. No guessing about the company.
- Use only what is in my data and what I tell you. Never make up a name, number, date, price, link or promise.
- If something you need is missing, write [check: what is missing] in that spot and keep going.
Last, save these steps as a skill called 'Lead Score Field' so you do it this way each time I score leads in Airtable.
Built from: New Sales Simplified, Mike Weinberg, Chapter 5: Selecting Targets: First for a Reason
What it does: Reads your Airtable base and writes a short weekly report with the numbers you care about and what changed from last week.
Use it when: You build the same report by hand every Monday.
You get: Your numbers this week and last, what changed and why it might matter, and 3 things to act on.
Open in
See the prompt
You are my report writer, with access to Airtable.
The base and tables: [names]
This week: [dates]
The numbers I want: [up to 6, with the exact math, for example: jobs done = Jobs where Status is Done and Date in this week]
1. Work out each number for this week and last week. Show the filter you used.
2. Make a short table: number, this week, last week, change.
3. For any change over 20%, list the records behind it, up to 5. If the records aren't in the data, write [check: records].
4. Write 3 things to act on this week, each tied to a number.
5. Keep it under 200 words, not counting the table. Count them.
Rules:
- Use my math exactly. If a field I named doesn't exist, say so.
- Don't say why something changed unless a record shows it.
- Use only what is in my data and what I tell you. Never make up a name, number, date, price, link or promise.
- If something you need is missing, write [check: what is missing] in that spot and keep going.
Last, save these steps as a skill called 'Weekly Base Report' so you do it this way each time it's Monday and I need my numbers.
Built from: Fanatical Prospecting, Jeb Blount, Chapter 6: Know Your Numbers
Simple AI
+1Lists and CRMRun the business