For developers

Cold-email deliverability that lives in your codebase.

UI-first agency tools assume a human clicking through a campaign builder. If you are shipping outbound from your own stack or letting an agent send on your behalf, you need endpoints and a stable contract — not a dashboard to screenshot.

The problem

What makes this hard today.

01

Agency tools do not fit a codebase

Sequencer UIs are built for a human operator, not a CI job or an agent loop. Their APIs are usually an afterthought bolted onto the UI, thin and under-documented.

02

Warmup and reputation are unwanted ops work

Standing up warmed mailboxes, watching Postmaster Tools, and pausing a degrading domain by hand is infrastructure work no one wants to own alongside the actual product.

03

Agents need real tools, not screenshots

An AI agent drafting outbound cannot click a UI. It needs a callable surface it can actually invoke — with guardrails that hold even when the agent is making the call.

How CogniLead helps

The mechanism, step by step.

  1. A REST API built for programmatic use

    POST /api/v1/leads to bring a lead in, then a clean send surface — OpenAPI 3.1 spec, Bearer pk_live_ keys, RBAC. See the full contract at /api/v1/docs.

  2. An MCP server for agent-driven outbound

    @cognilead/mcp-server ships 16 tools — 4 keyless (draft, score, check deliverability, health) so an agent can learn the surface before you commit a key, 12 gated for the real pipeline. See /mcp.

  3. Per-send pricing, not a seat tax

    $0.035 per send past the free 100/month. You pay for what you dispatch, whether that is one integration test or ten thousand real sends.

FAQ

Questions about Developers & AI-GTM builders.

Is there a UI at all, or is this API-only?+

There is a dashboard for keys, billing, and monitoring, but the send path itself is built API-first — the dashboard reads the same state your integration writes, it is not the primary interface.

Can an AI agent send through CogniLead directly?+

Yes, via the MCP server. Four tools work with no API key so an agent can draft, score, and check deliverability immediately; the twelve gated tools (ingest_lead, run outbound, suppress, and read endpoints) unlock with a key.

What does the API actually guard against?+

Every send passes suppression and reputation checks server-side before dispatch — an integration cannot accidentally bypass the circuit-breaker or re-mail a suppressed address, whether the call came from your backend or an agent.

How is this priced for a dev-heavy team?+

Per send ($0.035 past the free 100/month), not per seat — so wiring it into a CI pipeline, a staging environment, and production does not multiply your bill by headcount.

Related reading

CogniLead for developers — a deliverability API, not a UI