Cortex runs every AI call in the suite. The developer tier opens API access for agent runs, the agent catalog, per-agent configuration, cost tracking, and MCP integration.
Waitlist opens with launch. Check back when the page flips to Available.
Every job-scoring, proposal-drafting, and humanization run inside Prospector goes through Cortex. One agent platform; products supply the tools.
No completions vs agents split. The same call shape scales from a one-shot text rewrite to a multi-step loop with retrieval, humanization, and approval.
Every run carries tenantId, userId, agentType, and a correlationId. Cortex emits structured events; the Cortex admin panel reviews them.
Anthropic, OpenRouter, Google. Model selection lives in agent config; switching is a config change, not a refactor.
POST agents.run from anywhere with a tenant token. Run id back immediately; events streamed via Server-Sent Events.
Per-agent-type approval gates. Long-running loops pause and emit an approval_required event for the operator or end user to ratify.
One dashboard for token spend per agent type per tenant. No surprise invoices.
Cortex agents can call out to MCP servers (including Vault) as tools. Tool registrations live in agent config.
Cortex runs your agents. Vault holds their credentials, with leasing bound to the agent run. Synapse provides their long-term memory, with tenant scoping injected at the tool layer so the agent's prompt cannot bypass it.
The Cortex developer tier is not available yet. Leave your email and we'll let you know when it opens.
