Claude Integration
Connect your own Claude (Anthropic) account and pay the provider only for actual usage. Generate replies to buyer messages with one click. The AI uses your company context: the rules, tone and information you define yourself. You can also configure an automatic action that replies to messages with AI on its own.
What the Claude integration covers
You connect the integration with an API key from Anthropic Console and pay the provider directly for the replies. When adding it we send Claude a short test request - if the key or model does not work, you see an error and nothing is saved.
You decide which model writes the replies - you can choose versions of Claude Sonnet, Opus and Haiku. With one key you can add several integrations, for example with a different model or prompt for different channels.
In the System prompt field you describe who you are, what tone to use, opening hours, return rules and the reply language - it is the AI's only source of knowledge about your business. Max tokens caps the reply length, 1000 by default.
Claude replies to customers in the Send AI message to thread step of message actions. The reply goes to the customer right away, without approval, so you set the Message received from customer condition in the action - that way the AI does not reply to its own messages.
Claude gets the text of the latest message in the thread and your system prompt. It does not read the earlier conversation history or order data, so everything it should know about your business goes into the prompt.
How it works
Add the integration in the Ritevo panel and go through a short authorization with the provider. The process looks the same for every integration.
Each feature has its own switch in the integration settings, e.g. order syncing, message syncing or product synchronization.
Add automatic actions: documents, shipments and replies will run on their own.
Frequently asked questions
Yes, you provide your own Anthropic API key and pay the provider only for actual usage.
You define your company context: rules, tone and information about your offer. The AI uses it in every reply.
When Claude rejects the key while writing a reply, for example after the key was deleted, the integration needs reconnecting and the AI does not reply to customers. You create a new key in Anthropic Console and paste it with Update token - the settings stay unchanged.
No. Claude replies only through an automatic action and does not save replies as drafts for approval. When the provider returns an error, nothing is sent and you see the error in the message audit.