NOT ANOTHER
PLATFORM.
You want lean data plus a real API, not another platform to migrate onto next year. CMP42 is deliberately small, deliberately obvious, deliberately headless.
THE ASSUMPTION CHANGED.
HubSpot, Salesforce and Pipedrive were designed around an assumption that made complete sense when they were built: a human would open the UI. Every field, every pipeline view and every dashboard exists because someone was going to click on it. For teams that work that way, a UI-first CRM remains a sound choice.
What changed is the workflow. A sales manager who once opened the CRM to check the pipeline now asks an agent to summarise it — and opens the CRM only when the summary looks off. When the agent reads a schema designed for people clicking through screens, it struggles: it invents fields, fabricates stages, fills gaps with guesses.
If that describes your team, the question is not which platform to move to next. It is whether you need a platform at all — or a well-structured graph and a boring, fast API that every agent can use.
LEAN DATA. A REAL API.
Lean defaults
Contact, company, deal, task, activity — with sensible, small defaults. Extension is explicit: if a field is not in your ontology, it is not on your contact. Lean defaults →
MCP, REST and GraphQL
Every model becomes MCP tools automatically — in the style of get_contact or list_deals_at_stage. Every operation is also exposed through REST and GraphQL, with the same auth and audit trail. API →
Your ontology in git
Models are declared in TypeScript-shaped YAML, reviewed like code and migrated with one command. Your schema is a product artefact you own, not a configuration spread across settings pages. Models →
Your hosting choice
Managed by us in the EU, or run yourself on your Postgres and object store. Same binary, same models. Your data, your call. Self-hosting →
FOUR STEPS, CONCEPTUALLY.
Every CRM move is different. This is the shape of one onto CMP42 — what happens in which order, and why.
- Inventory what you actually use. Long-lived CRM instances accumulate properties, pipelines and objects nobody reads any more. Before moving anything, list the objects, fields and relationships your team — and your agents — genuinely rely on. That list is the first draft of your ontology.
- Map your objects to Models. Contacts, companies and deals map onto CMP42’s lean defaults. The custom objects and fields that survived your inventory become explicit model declarations, versioned in git and reviewed like code. Relationships you used to encode in free-text fields become typed edges. Design Partners do this in a first-model mapping session during onboarding week.
- Import. Bring the records across. Because every write is validated against your Models, malformed data surfaces as clear errors instead of silently becoming part of the graph an agent will read.
- Point your agents at MCP. Connect Claude Desktop, Claude Code, Cursor, Cline or your own agent on the MCP SDK to the generated tools. Point the rest of your stack at REST or GraphQL. Your team keeps working where it already works — in its agents, in Slack, in the apps you build.
In the Alpha, CSV import for the standard objects — contacts, companies, deals and notes — ships as a built-in mapper, and everything beyond it (custom properties, historical timelines, attachments) is migrated with founder support at no extra charge as part of the Design Partner Program. The one-click HubSpot API connector with OAuth arrives with the Beta in Q1 2027; until then, working directly with the founder tends to be faster than a generic connector would be for the shape of migration Design Partners actually run.
The sketch illustrates the mapping idea — legacy objects on the left, CMP42 models on the right, and a deliberate decision about everything else. It is not an importer configuration.
# illustrative — the mapping idea, not a tool
contact ← contacts # lean default, extend explicitly
company ← companies
deal ← deals # stages declared in your model
note ← call notes # linked by typed edges
doc ← proposals, briefs
# everything else: keep it only if a reader uses itIS CMP42 RIGHT FOR YOUR MOVE?
The founder has spent fifteen years wiring companies to HubSpot, Salesforce, Pipedrive and the CRMs before those. The honest lesson: the best CRM is the one that fits how your team actually works. Here is when CMP42 fits — and when it does not, yet.
A good fit
- Your team reaches CRM data mainly through agents — or wants to.
- You want your ontology in git, reviewed like code.
- You have engineers, or a partner, who are comfortable with an API.
- You need EU hosting or self-hosting for your customer data.
Not a good fit (yet)
- Your team needs a finished UI for daily pipeline work. CMP42 has no forced UI; the admin surface is a data table and a migrations CLI.
- You expect dashboards out of the box. There are none until you write them yourself.
- You need a production system today. CMP42 is in Alpha with weekly releases — things will break.
- You cannot run a real workload during Alpha. Then wait for Beta, which opens in Q1 2027 with a waitlist.
Bring your CRM mess to the call
The application asks what CRM or spreadsheet you are on today and what the messiest part of your CRM-to-AI workflow is. Fifteen minutes, no slide deck.
→ Apply to be a Design Partner