Case study: built deeper than the directory

A custom Pipedrive read-write connector for Claude

A sales team had outgrown the directory connector for their heavily customized Pipedrive. We built one that lets Claude create and update deals, activities, and custom fields with scoped write.

Build my connector Pipedrive connector

How do you connect a heavily customized Pipedrive to Claude?

Full read-write into a custom Pipedrive setup means building a custom MCP connector that writes to your real schema, including custom fields, not just the standard ones. We built one for an anonymized sales team so Claude creates and updates deals, activities, and custom fields with scoped, gated write controls.

The situation

The team (anonymized at their request) ran their entire sales motion in Pipedrive, but years of tuning had left it heavily customized: extra pipelines, required custom fields, and conventions that the standard objects did not capture. The directory connector could read most of this, but it could not reliably write back to the custom fields their reporting and routing depended on. So Claude could summarize a deal but not actually move it forward, which meant reps still did the data entry by hand.

What we built

We built and hosted a custom MCP connector mapped to this account's actual Pipedrive schema, with read and write tools kept separate. It authenticates as each user's own Pipedrive account, so it inherits their existing permissions and never grants Claude more access than the login already has. Create and update flow freely inside a conversation a rep starts, while deletes are gated behind a human-in-the-loop confirmation. As the team adds or renames custom fields, we update the connector so it keeps matching their setup.

How does Claude write back to custom Pipedrive fields?

Writing back to custom Pipedrive fields means the connector maps each custom field, then creates and updates deals, activities, people, and organizations through the user's own Pipedrive login. Create and update flow freely inside a conversation a rep starts, while deletes stay gated behind a confirmation step.

Create new deals and update existing ones, including stage, value, owner, and the team's custom fields.
Log and update activities (calls, meetings, and tasks) against the right deal, person, and organization.
Read and write the custom fields that drive the team's pipeline logic, not just the standard Pipedrive fields.
Find and update people and organizations so deal records stay linked to the right contacts.
Summarize a deal's full history, notes, and recent activity before suggesting the next step.
Gate destructive actions: deleting a deal or activity requires a human-in-the-loop confirmation the team controls.

The outcome

With write access into the custom fields that mattered, the team moved from reading deals in Claude to actually updating them there. Reps now ask Claude to log a call, advance a deal, and fill the required custom fields in one step, instead of switching back to the Pipedrive UI for every change.

Data entry: Deals, activities, and custom fields update from the conversation, not by hand in the Pipedrive UI.
Custom fields: The customized fields and stages the directory connector skipped are written correctly, not left blank.
Pipeline hygiene: Calls and follow-ups are logged as the work happens, so the pipeline stays current without a separate cleanup pass.

Want the same depth for your Pipedrive? See what a full read-write Pipedrive connector covers, including how scoped write and gated deletes work.

Frequently asked questions

Why build a custom Pipedrive connector instead of using the directory one?

The directory connector was too shallow for a heavily customized Pipedrive. It could not reliably write to the custom fields the team's pipeline logic depends on. We built a deeper connector with full read-write across deals, activities, and those custom fields, then maintain it as the account's schema changes.

Can Claude change live deals in this setup?

Yes. Claude creates and updates deals, activities, and custom fields inside a conversation a team member starts. Create and update flow freely so work moves forward, while deletes are gated behind a confirmation step. The connector authenticates as the user's own Pipedrive account, so it never grants more access than that account already has.

Does the connector see more of Pipedrive than the user does?

No. The connector inherits the permissions of the Pipedrive account it authenticates as. A rep sees and edits only what their own login allows, and an admin's session reflects admin scope. Claude cannot reach pipelines, deals, or fields the underlying account cannot already reach.

What happens when the team adds or renames custom fields?

We update the connector so it keeps matching the live schema. When the team adds, renames, or retires a custom field, we remap the affected read and write tools so Claude keeps writing to the correct fields. The connector is maintained as the account evolves, not built once and left to drift.

Tell us what you run. We'll build the connector.

Pick your tool, choose what Claude should do, and get an instant scope and price. No call required.

Build my connector Talk to us