AI providers

Autocloz + Anthropic (Claude)

Anthropic's Claude models can run Autocloz's AI features when you add your own Anthropic API key under Settings → AI & Models and choose Anthropic as the active provider. Claude then writes email drafts, personalizes messages per lead, classifies inbound replies, drafts responses for you to review, and powers AI lead lists. The model list offered includes claude-opus-4-6, claude-sonnet-4-6 and claude-haiku-4-5. In bring-your-own-key mode the tokens bill to your Anthropic account, not to Autocloz credits.

Autocloz and Anthropic (Claude) integration

Bring your own accountWorkspace-scoped, no cross-org leakFree-forever CRM core

Start free See pricing

What you can do with Anthropic (Claude)

Anthropic (Claude) plugs into the AI replies & summaries stack — bring your own api key, scoped to your workspace.

Drafts you approve

Email drafts, per-lead personalization and reply drafts are generated on request and wait for you. Autocloz does not send AI replies on its own.

Reply classification

Every inbound reply can be bucketed as interested, not interested, out of office and so on, which drives follow-up stops and reporting.

Model choice per workspace

Pick a Claude model from the list Autocloz fetches from Anthropic with your key. If that fetch fails, a built-in fallback list is shown.

How to connect Anthropic (Claude)

Bring your own API key — you keep control of the account, the spend and the data boundary.

1
Create an Anthropic API key
In the Anthropic Console create a key for the workspace whose billing you want AI usage to hit.
2
Open Settings → AI & Models
Choose Anthropic as the provider, paste the key, and pick a model. The key is encrypted and only its masked form is shown back.
3
Turn features on or off
Each AI feature has its own switch, and channel personalization can be switched per channel. Leave off anything you do not want running.
4
Try one draft
Generate an email draft in a campaign step or a reply draft in the Unibox and check the tone before you rely on it.

Picking a Claude model for each job

Classification and CSV mapping are short, repetitive tasks, so a small, fast model is enough and keeps cost down. Drafting a first email or rewriting one for a senior prospect benefits from a larger model. Autocloz uses one active model per workspace, so choose the model that suits the feature you use most, and switch when your usage changes.

Which Autocloz features Claude runs

Autocloz has ten AI features, and all of them go to the active provider: email drafting, email personalization, reply classification, reply drafting, lead extraction from pasted text, prospecting enrichment (seniority, department and an ICP-fit score), channel personalization for SMS, WhatsApp, LinkedIn and call scripts, AI lead lists from a plain-English ICP, chatbot answers from your knowledge base, and smart CSV column mapping.

Reply classification runs on every inbound reply, so its volume follows what prospects send. Chatbot answers run on a public page, so anyone visiting your site can trigger them. Watch those two if you care about Anthropic spend. More on the reply side in AI replies.

Your key versus Autocloz credits

A workspace runs AI in one of two billing modes. In credits mode, Autocloz calls the model and deducts credits after a successful response. In bring-your-own-key mode, calls go out on your Anthropic key, Autocloz deducts nothing, and it writes a zero-cost usage row with the real token counts and an estimated dollar figure so you can see what each feature is costing you. Credits are never charged for a failed call in either mode. Pricing lists what each plan includes.

What Anthropic (Claude) does not do in Autocloz

It does not send anything by itself. Drafts wait for a person.
It does not run AI voice calls. Voice-agent calls use Sarvam over Telnyx, not Claude.
The old blurb for this page mentioned an "analyst mode" for drilling into leads. No such feature exists in Autocloz.

Troubleshooting Anthropic (Claude)

AI actions return byok.no_key.: The workspace is in bring-your-own-key mode and the active provider has no key. Add the Anthropic key or switch provider.
An AI action returns ai.feature_disabled.: That feature is switched off in Settings → AI & Models.
The model list looks out of date.: Autocloz could not reach Anthropic's models endpoint with your key and showed its fallback list. Check the key is valid.

Part of the AI replies & summaries stack

Anthropic (Claude) is one of the AI providers you can bring to power replies, summaries and assist features. They plug into the same AI surfaces — you choose which key drives them.

Explore the AI replies & summaries feature