A CRM built as infrastructure, with AI as the interface.
Trickjar separates where your customer data lives from how you work with it. Four layers, each one replaceable without moving your data.
Typed tools, scoped to one workspace.
Models never get raw database access. They call documented tools, and every call is limited to the workspace whose key it uses.
| Tool | What it does | Example prompt |
|---|---|---|
| find | Search contacts, companies and deals by any field, list or activity window | "Which open deals have had no activity in two weeks?" |
| update | Create or change records, fields, owners and stages | "Move Acme to negotiation and set the close date to the 30th." |
| enroll | Add contacts to a sequence or a LinkedIn task queue | "Enroll every new CFO from this week's list in the finance sequence." |
| report | Aggregate pipeline, activity and sequence results | "Pipeline by stage and owner, this quarter versus last." |
Isolation by design, not by filter.
A database per workspace
Brands and clients never share tables. Each workspace is its own database.
Nightly backups
Every workspace database is backed up every night.
Full export
Export everything, whenever you want, in an open format.
Host it where you want
Basic storage is included. Keep the rest on our hosting, in your own cloud or on-prem.
Headless, not blind.
When you want to click, the web app covers the daily work.
Lists and saved views
Filter, save, share.
Sequences
From your own domain and mailboxes.
LinkedIn task queue
Manual steps, tracked.
Activity timeline
Every touch on one record.