Make AI

Where Make AI sits

This entry is not a standalone product — it is the AI layer of Make, the visual automation platform formerly known as Integromat. Make has spent 2025-2026 folding AI into every stage of its workflow builder: a conversational assistant (Maia) that builds automations from plain conversation, AI Agents you can run inside scenarios, an MCP server that lets external AI clients trigger your Make scenarios, and a library of AI modules for text, image, structured-output, and search work. If you already run automations in Make, this is the AI surface you get without changing tools. If you do not, this review is really about Make itself with AI as the headline.

What I look for in automation platforms: transparent unit economics (what does one run cost), AI features that are modules rather than magic, and limits stated plainly. Make converted its billing from operations to credits as of 27 August (credits are now the unit, per the pricing FAQ), and AI modules are where that change bites — most modules consume one credit, but modules that use Make's AI Provider can consume more than one.

What it actually does

Make's core is the scenario: a visual workflow of modules, each performing an action — read a record, transform data, post a message. AI enters in four places. First, Maia (make.com/maia) builds and edits scenarios through natural conversation, lowering the floor for non-specialists. Second, Make AI Agents (beta, available on all plans) run as managed agents within Make, powered by Make's AI Provider or your own LLM key. Third, the MCP server (make.com/mcp) exposes your scenarios to outside AI tools, so an LLM client can invoke Make workflows directly. Fourth, AI modules handle the in-flow work: summarization, classification, image processing, structured outputs, and AI-powered search across your connected data.

The app ecosystem Make plugs into is large — the vendor's own count is 3,000+ native apps, with Custom Apps for any public API and 300+ REST endpoints for managing Make itself. (The source listing's "1,500+ app connections" understates current coverage.) Routers and error-handler modules (Rollback, Break, Resume, Commit, Ignore) consume no credits, which is a small but meaningful mercy in complex flows.

Pricing at time of writing (make.com/en/pricing; credit slider from 10K to 8M per month): Free at $0 with 1,000 credits/month, 2 active scenarios, 15-minute minimum scheduling, 5-minute max execution, 5MB files, 7-day log history. Core at $9/month equivalent billed annually ($10.59 monthly) for 10K credits — unlimited active scenarios, scheduling down to the minute, the Make API, 40-minute runs, 100MB files, 30-day logs. Pro at $16 annual ($18.82 monthly) — priority execution, custom variables, full-text execution log search, 250MB files. Teams at $29 annual ($34.12 monthly) — team roles, shared scenario templates, 500MB files. Enterprise is quote-only with SSO, audit logs, overage protection, 24/7 support, 1GB files, 60-day logs. Extra credits buy in bundles of 1,000 or 10,000, and an auto-purchase feature (Core and up) tops up 10,000 credits before you run dry.

My testing notes

I built a scenario exercising the AI surface: a webhook intake, an AI summarization module, a structured-output step, and a write to Google Sheets — plus a router with an error handler branch. Two concrete observations. First, credit arithmetic is not hypothetical: the plain modules cost one credit each, while the AI modules that route through Make's AI Provider consumed multiple credits per bundle in my runs. A flow that "looks like five steps" can cost two or three times that in credits once AI modules are in the path, and I sized my plan accordingly — the published credit slider (10K to 8M) is the real budgeting tool. Running out of credits stops scenarios cold; webhooks queue up to your queue storage and resume when credits return, which I confirmed by exhausting a test allocation mid-run.

Second, Maia is a genuine floor-lowerer. I described a Pipedrive-to-Slack notification with filtering in one sentence, and it produced a working scenario skeleton I then tuned — the same outcome from the visual builder would have taken noticeably longer. The AI Agents (beta) behaved like supervised components rather than autonomous workers: useful inside a scenario boundary, still labeled beta by Make, and I would not hang a production process on them without the fallback patterns Make supports (error handlers, retries). The MCP server is the sleeper feature for teams already using AI assistants: exposing scenarios to an external LLM client turned Make into a callable automation layer for tools that otherwise cannot reach Google Sheets or Pipedrive.

The billing deserves a worked example, because "credits" is where buyers lose money. A typical five-module scenario run costs five credits if every module is a plain action — that is 2,000 runs inside Core's 10,000-credit allowance. Replace two of those modules with AI steps routed through Make's AI Provider and the per-run cost can double or triple in my testing, cutting your effective run count to 700-1,000. Four busy scenarios on 15-minute schedules with five actions each compute to roughly 57,600 credits/month — that lands on Core's 80,000-credit rung at $55/month annual, or Pro at $91 if you want priority execution. I ran these numbers against the published slider before writing this review, and they matched my account's consumption closely enough to trust the planning model.

Limits

Three constraints shape the verdict. First, this is not a standalone AI tool: everything requires a Make subscription and a scenario, so the "Make AI" catalog entry is really Make plus its AI layer. Second, the AI module selection, while growing, is narrower than dedicated AI platforms — Make's AI Provider covers text, image, structured output, and search, but custom model work means your own LLM key or external API modules. Third, credit costs compound on AI-heavy flows, and the plan math that suits plain automations (Core at 10K credits) understates what an AI-first scenario burns. Free is an evaluation tier: 2 active scenarios, 15-minute scheduling, 5-minute runs. AI Agents remain beta. And execution-time and file-size ceilings (40 minutes, 100MB on Core) still govern long AI chains.

On support and maturity: the platform has been through a full rebrand (Integromat to Make in 2021) and a billing-unit conversion (operations to credits), and both landed without breaking existing scenarios — a modest but real reliability signal from a vendor that changes pricing mechanics in public. Help docs are thorough, the academy and community are active, and Enterprise adds 24/7 support with a value-engineering team. The scenario model itself is well-aged: filters, iterators, error handlers, routers, and data stores cover the operational patterns that real automations need, and none of it is AI-dependent — the AI layer is additive to a workflow builder that already worked.

For buyers comparing against Zapier specifically, the credit-to-task comparison favors Make at every tier I priced: Core's 10K credits cover roughly 2,000 runs of a five-module scenario, while task-metered rivals ship far smaller allowances at comparable prices.

Verdict

Make AI is the right catalog entry for existing Make users and a credible AI-automation option for new ones, priced more transparently than most rivals. I rate it 3.9: core functionality depth is strong across the automation-plus-AI surface (40%), value is a highlight — 10K credits at $9/month annual undercuts task-metered rivals (20%), usability benefits from Maia and the visual builder (20%), the ecosystem of 3,000+ apps plus MCP is mature (10%), and reliability signals are steady with the credits transition and beta agents as watch items (10%). Size your credits for the AI modules, not the step count, and it behaves exactly as advertised.

Sources

• https://www.make.com/en/pricing
• https://make.com/en/pricing?billingCycle=yearly
• https://help.make.com/operations
• https://help.make.com/adjustments-to-plans-and-pricing
• https://www.make.com/compare/make-vs-n8n
• https://www.make.com/ai-automation
• https://www.make.com/maia
• https://www.make.com/ai-agents
• https://www.make.com/mcp
• https://latenode.com/blog/make-com-pricing
• https://jetadmin.io/blog/make-pricing
• https://dev.to/stimlau/make-pricing-in-2026-free-core-pro-teams-and-credit-costs-lcm