cf
CallFunnel.ai

Features

A control plane, not a chatbot.

CallFunnel is the layer above the LLM, the channel, and the CRM. You write the policy. We run it everywhere a customer might reach you, and we keep one memory across all of it. Here's the surface area.

Pre-launch simulation

Test your campaign against simulated calls before you dial a real customer. The agent faces the objections it'll actually hit, the weak points surface, and the knowledge bank gets smarter each round.

New · See how it works →
NL

Plain-English rule books

Describe the policy. Opus drafts the script and the safety rails. Validators catch the gaps. Self-heals on the second pass.

How authoring works →

One timeline per customer

Every call lands in the same Conversation — same context, same tone, same memory of what already happened. The AI never restarts cold.

How memory works →
📄

Knowledge bank

Drop in docs, FAQs, policies, scripts. Pull from Shopify, REST, SQL, CSV. Voyage embeddings, Qdrant retrieval, PII redacted before the LLM.

KB & data sources →

Approval-in-the-loop

Auto-approve under a threshold. Drop above-threshold offers into Slack with the transcript. Escalate to a backup approver if nobody answers. The AI never invents a discount.

Approval ladder →

Pre-integrations

Twilio, Exotel, Deepgram, Cartesia, Claude, Elastic Email, Slack, Keycloak, Vault — already wired. Bring your own keys or use ours.

Full catalogue →

Hosted MCP

CallFunnel exposes its workflows, knowledge bank, and timeline as MCP tools. Drive it from Claude.ai, Cursor, or your own agent.

MCP endpoint →
🔑

BYOK or bundled

Bring your own Anthropic, Twilio, Cartesia, Deepgram keys — pay $150 / ₹10,000 flat per month. Or use ours at ₹9 / $0.10 per minute. Switch any time.

BYOK setup →
🛡

Compliance & security

DPDP / TCPA / GDPR scaffolded. Hard tenant isolation. DSR export & erase under 30 seconds. PII redacted pre-LLM. Audit logs per tenant.

Trust page →