How to build a CRM with AI and Airtable.
I use this to build and shape the CRM itself by talking to it - not to maintain an existing one (that is the clean-up skill) or run it on a schedule (that is the routines skill), but to design and stand up the base in the first place, and to keep reshaping it as the business changes. Airtable is the CRM; Claude is how I build it. I describe the pipeline I want in plain English and it creates the tables, fields, links and views. I reach for this when starting a CRM from scratch, adding a new object (a new pipeline, a partner table, a placements log), or fixing a structure that no longer fits.
How I use this, and what for
I use this to build and shape the CRM itself by talking to it - not to maintain an existing one (that is the clean-up skill) or run it on a schedule (that is the routines skill), but to design and stand up the base in the first place, and to keep reshaping it as the business changes. Airtable is the CRM; Claude is how I build it. I describe the pipeline I want in plain English and it creates the tables, fields, links and views. I reach for this when starting a CRM from scratch, adding a new object (a new pipeline, a partner table, a placements log), or fixing a structure that no longer fits.
What the AI actually does
You describe the CRM you want. The AI reads what is already there, designs the schema, and creates the tables and fields with the right types and relationships. Restructures that feel daunting - splitting contacts from companies, adding a linked activities table - become a paragraph. The manual version is an afternoon clicking through Airtable's interface. This is one prompt.
Airtable is a no-code platform that pairs a spreadsheet interface with a relational database, so it is built for exactly this: linked records that connect data across tables and stay in sync, rich field types, and multiple views (grid, kanban, calendar) over the same data. That is why AI can build a working CRM in it from a description rather than wiring up a database by hand.
The tools and APIs you need
The right prompts
Stand up the CRM
```
Build me a sales CRM in Airtable. Tables: Companies, Contacts, Opportunities,
Activities. Link Contacts to Companies, Opportunities to both. Opportunity
fields: stage (single-select: New, Discovery, Proposal, Negotiation, Won,
Lost), value, owner, next step, next-step date, close date. Activities:
type, date, linked opportunity, notes. Check the existing base first and
extend it, do not duplicate what is there.
```
Add an object later
```
Add a Placements table to the ATS base, linked to Companies and Contacts,
with fields for role, candidate, fee, start date and status. Then make a
view of placements closing this quarter.
```
How to do it
Describe the CRM you want
in plain English: the objects, the key fields, how they link, the stages a deal moves through.
Let it design the schema
- tables for companies, contacts, opportunities and activities, with linked-record relationships and sensible field types (single-select for stage, dates for next-step, links between contact and company).
Have it create the tables and fields
in Airtable. It checks the existing base first so it extends rather than clobbers.
Add the views you work from
- my pipeline, deals closing this month, stale deals, by owner.
Seed and test
with a few real records to confirm the structure holds. The clean-up skill then keeps it current; the routines skill maintains it on a schedule.
Reshape when the motion changes
- a new pipeline or object is another short prompt, not a rebuild.
See the difference
Doing it by hand: build a crm with ai and airtable the manual way - slow, and the first thing to slip when you are busy.
With AI: you describe what you want in plain English and it does the work, on-brand, in minutes.
Let AI carry the heavy lifting; you keep the judgement and the final say.
How you'll know it's working
- A CRM that fits how you actually sell, built from a description, not a template.
- Tables linked properly - contacts to companies, deals to both - so the data joins up.
- New objects added in a prompt, without an Airtable rebuild.
- A clean structure the clean-up and routines skills can then keep current on their own.
Questions people ask
How do I build a CRM with AI?
You describe the pipeline you want in plain English - the objects, the fields, how they link, the deal stages - and the AI creates the tables, fields, links and views in [Airtable](https://www.airtable.com/) for you. It reads the existing base first, so it extends what is there instead of duplicating it.
Why Airtable for an AI-built CRM?
[Airtable](https://www.airtable.com/) is a no-code database and app platform with linked records, rich field types and multiple views, all reachable through its API. That structure is what lets AI stand up a real, relational CRM from a sentence rather than a flat spreadsheet.
How is this different from keeping the CRM clean or running routines?
This skill builds and reshapes the structure of the CRM. The clean-up skill (ai-crm-and-pipeline) keeps an existing base accurate by logging and reporting in plain English. The routines skill (ai-crm-routines-automation) runs that upkeep on a schedule without you. Build first, then keep clean, then automate.
Hire with Assessment.
£7-10k flat fee. The methodology, delivered.
See Hire with Assessment