We connect assistant and automation features in your system to Anthropic's Claude models through the Anthropic API.
Claude models handle long documents well, which suits work such as reviewing contracts, summarising case files or extracting details from lengthy reports. Built into your software, they work on your own records, within the screens your team already uses, and only with the data you choose to share.
What we build with it
- Document analysis and summarisation
- Drafting assistance
- Natural-language questions over business data
- Structured extraction from long documents
How it works
- A user selects a document or asks a question in the system.
- The system prepares the request, including only the records and text that user may see and the task needs.
- It sends the request to the Anthropic API from the server, never from the user's browser.
- The response is returned to the screen for the user to check, edit and then save or discard.
Data exchanged
- Prompt content including any business context supplied
- Generated response
- Token usage
What is needed to set it up
- An Anthropic API account
- API key configured server-side
- Agreement on what data may be sent
Good to know
- Anthropic bills usage to your own account according to the amount of text processed, at rates set by Anthropic for each model.
- Anthropic's commercial terms describe how API data is handled and retained. Check the current terms against your own obligations to clients.
- Output can contain mistakes, so we design features where a person reviews results before they are relied on.
