Make.com has shipped a built-in Anthropic Claude connector, adding native integration with Claude models directly to the platform. The connector was announced in Make's March 24 release notes and is now available across all paid plans.
Until now, using Claude in Make.com scenarios required setting up the HTTP module with Anthropic's API directly — workable but verbose, with no native module surfacing Claude's specific capabilities. The new built-in connector changes that.
What's actually in the connector
The Claude connector exposes the standard Claude API capabilities as Make modules with proper input/output mapping. From the documentation, this includes:
- Messages API — single-turn and multi-turn conversations with Claude
- Tool use — Claude can call back to Make tools, enabling agent-like workflows
- Vision — image inputs for multimodal tasks (document analysis, screenshot interpretation)
- System prompts — proper handling of role-based prompting
- Model selection — choose between Claude Opus, Sonnet, and Haiku tiers
- Streaming — partial response support for long-running generations
The connector handles authentication via OAuth2 against Anthropic's API, removing the need to manage API tokens in HTTP module headers. It also surfaces Claude-specific parameters (max_tokens, temperature, top_p) as proper module inputs rather than buried in JSON.
Why this matters
Three angles where this changes things:
Lower entry barrier for Claude in Make.com. Previously you had to know the Anthropic API spec, format requests in HTTP modules, and parse JSON responses manually. Now it's drag-and-drop. For citizen developers and non-technical builders, this is the difference between "I might use Claude someday" and "I'm using Claude in tomorrow's workflow."
Native AI Agents integration. Make.com's AI Agents feature can now use Claude as the underlying model with proper connector support, instead of going through HTTP module workarounds. This means cleaner agent scenarios with better debugging and clearer cost tracking.
Multimodal workflows get easier. Sending an image to Claude for analysis was technically possible via HTTP modules but painful. With the native connector, "OCR this invoice with Claude" or "Describe this screenshot" become 2-step scenarios anyone can build.
How it compares to using OpenAI in Make.com
Make has had a built-in OpenAI connector for over a year. The Claude connector matches that level of integration, putting Claude on equal footing with GPT in the Make ecosystem. For teams choosing between models, this removes the "but OpenAI is easier to integrate" objection.
Differences worth noting:
- Claude's pricing model (input tokens cheaper, output tokens more expensive than equivalent GPT-4 models) means operations cost calculations differ slightly between providers
- Claude's strict output formatting (especially with XML tags and structured responses) tends to give better results for data extraction tasks where reliability matters
- Claude Haiku at sub-cent per request is now an accessible default for high-volume light tasks
Practical use cases that just got easier
From projects we've delivered using Claude (previously via HTTP module), here are the patterns that get simpler with the native connector:
Document data extraction. PDF or image in, structured JSON out. Claude's tool use makes this reliable. Pair with our custom integration service if you have specific document formats to handle.
Email triage and response drafting. Incoming email → Claude classifies urgency, sentiment, and suggested action → drafts a response in your tone. Faster than building this against the raw API.
Content rewriting for brand voice. Pass content + brand voice guidelines + transform target. Claude is particularly good at this category vs other models.
Long-form content summarization. Claude's larger context window makes it the default for summarizing 50-page documents, transcripts, or research reports.
Cost reality check
Native connector or not, AI calls in Make.com consume operations. Each Claude call costs:
- 5 operations for the module invocation (Make.com's standard AI module cost)
- Plus Anthropic's actual API charges, billed separately to your Anthropic account
If you're new to running AI workflows in Make.com, the operations consumption can surprise you. Budget accordingly. See our Make.com pricing guide for typical consumption patterns.
What I'd build first
If you have a Claude API key and a Make.com Pro plan, three scenarios are worth building this week:
- Email summary digest — daily digest of unread emails summarized by Claude
- Document extractor — drop PDF in Drive folder, get structured fields back in Airtable
- Slack copilot — Slack message → Claude → response in thread
Each takes 30-60 minutes to build now (vs 2-3 hours with the HTTP module before). Quick wins to learn the connector.
Strategic implications
This release confirms Make.com's approach to AI: be model-neutral, support all the major providers natively, and let users pick based on their use case. Compare to Zapier (still primarily OpenAI-focused) or n8n (DIY for everything).
For Anthropic, the integration is a meaningful distribution win — Make.com's user base of 500,000+ now gets one-click access to Claude in their automations.
Official connector documentation is in the Make.com app catalog.