The CRM your agents run
Scalar exists because agent work deserves a better home than a folder of markdown files. This is the why.
The problem
Agents are brilliant at doing. Terrible at remembering.
An agent researches a company, drafts the outreach, finds the right person, and then forgets all of it. The output lands in scattered markdown files and loose notes: no schema, no dedup, no interface, no consistency.
So it re-researches the same company twice. It contradicts itself. And you can't browse, query, or trust any of the work it already did. Agents without a real database are goldfish with PhDs.
The answer
Give the work a structured home.
Scalar is the place agent work lands and stays consistent: typed companies, contacts, deals, and emails in a real database, deduped and validated on the way in.
And it comes with a real UI, so you can see, trust, edit, and navigate everything your agent did. Structure for the machine, an interface for the human, intelligence built into both.
Agent-first
Built for agents to operate, over MCP.
Scalar was built agent-first from day one. A secure MCP surface and per-user API keys mean your agent, whether that's Claude, OpenClaw, Hermes, or the built-in one, operates the CRM directly: discovering leads, enriching records, tracking deals, running email.
Agents and humans go through the same operations layer, so there is never drift between what the agent does and what you see.
The principle
A wrong answer is worse than no answer
Enrichment must attach to the right person at the right company, verified, every time. If Scalar can't confirm a match, it returns nothing rather than a guess, and a miss is never charged. Accuracy beats coverage. That rule is the soul of the product: a CRM you can trust is the only kind worth having.
Read the manifestoConnect your agent. It just works.
Point your agent at Scalar over MCP, or use the built-in one, and watch your CRM fill itself.