Overview
Anthropic is now an available provider in BYOM (Bring Your Own Model). You can connect your own Claude models to Pipefy using your API key and choose which model each AI Agent uses. Companies that already run Claude internally can now power their Pipefy Agents with the same models, under their own credentials.
What's new?
- Anthropic as a BYOM provider: available in the provider carousel, in the Admin Panel, alongside the existing providers.
- 3 Claude models available: Haiku 4.5, Sonnet 4.6, and Opus 4.8.
- Setup by Org Admins: in the Admin Panel, just enter a Name, your API Key, and select the model version.
- Model selection per Agent Behavior: Pipe Admins choose which Claude model to use in each Behavior, based on what each workflow needs.
- Full configuration management: create, edit, and delete your Anthropic provider configuration whenever you need.
How does this help?
- Standardize your AI stack: if your company has adopted Claude, your Pipefy Agents run on the same models, with governance centralized in your own API key.
- The right model for each task: combine speed (Haiku 4.5), balance (Sonnet 4.6), or maximum reasoning capability (Opus 4.8) depending on the complexity of each Agent Behavior.
- Full control over your configuration: provider management stays with Org Admins, with the freedom to edit or replace the configuration at any time.
Use cases
1. Contract review Agent in the Legal workflow
The Agent responsible for reviewing contracts in the analysis phase of the Contracts pipe now runs on Opus 4.8, connected through the company's own API key. The Agent reads attachments, flags risk clauses, and fills in the review fields with detailed reasoning. The legal team stops manually triaging every document and only acts on flagged contracts.
2. Ticket triage Agent in the Support pipe
A high-volume request classification Agent runs on Haiku 4.5, prioritizing response speed. For every new card, the Agent categorizes the request type, sets the priority, and triggers the routing automation to the right phase. The support team stops classifying tickets manually and first-response SLA drops without raising cost per execution.
3. Expense review Agent in the Finance workflow
In the Reimbursements pipe, the Agent that validates receipts and categorizes expenses runs on Sonnet 4.6, balancing accuracy and cost. The Agent checks amounts, identifies the expense category, and flags mismatches between the receipt and the requested value. In-policy approvals move forward without human intervention and Finance reviews only the exceptions.
4. AI governance for companies standardized on Claude
The IT team of a company already using Claude in other tools connects the same API key to Pipefy. All AI consumption from Agents runs through the corporate Anthropic account, with usage visibility and compliance with internal AI policies. The company keeps a single model vendor relationship, with no parallel stacks to manage.
Most relevant for
- Org Admins
- Pipe Admins
- IT Leaders
- Teams already using Claude internally
- AI Agent builders
Plans and availability
This feature is being rolled out gradually to all accounts.
Documentation
For more details check the Help Center article.
