.DkNUKaKx_ZLAT4m.webp?dpl=dpl_fd9buTzZJnrTBKP2HfWPqBq4zdMb){kind=small}
The real difference
Make is a capable tool for technical individuals and small teams automating lightweight processes. Its visual scenario builder is accessible, the app library is broad, and for straightforward trigger-action workflows it works well. It has also moved into AI: Make AI Agents (still labelled beta), a hosted MCP server, and Maia, an assistant that builds scenarios from a plain-language request. Make is now part of Celonis, which describes itself as a process mining and execution management company valued at $13 billion.
The ceiling appears at enterprise scale, and Make publishes one edge of it. Its pricing page, read on 2026-10-08, lists the “maximum run time of a single scenario execution” as 40 minutes on every paid plan, Enterprise included, and 5 minutes on Free. If a long migration or a bulk backfill is on your roadmap, that is the number to design around.
Credit metering is the other shape to understand. Most actions consume one credit and some features that use Make’s AI provider may use more, so the bill follows volume rather than the value of the workflow.
The pattern teams follow: start on Make, outgrow it once workflows get long-running or governance requirements appear, and move to a platform built for that environment. Tray.ai is where that path leads.
Where Make wins
Accessibility and published price for lightweight use. Make has a free tier and publishes its rates, starting at $12 a month for 10,000 credits, which most platforms at this end of the market do not. The scenario builder has a shallow learning curve. For a 10-person team connecting two SaaS tools, Make is a reasonable starting point.
It is also no longer a standalone bet. Being part of Celonis puts a large, well-funded parent behind the product, which answers the vendor-longevity question a procurement team would otherwise raise.
Make’s app library is broad: 3,000+ apps by Make’s own count. If the app you need is in it and your workflow is simple, Make ships it fast. And Make is not standing still on AI. Agents are built on the same canvas as scenarios, with a reasoning panel that shows each decision, and the Make MCP Server lets Claude or ChatGPT run your scenarios and, on paid plans, edit them. If you mainly want an AI assistant to trigger existing automations, Make covers that.
Where Tray.ai wins
- Scale without a credit meter. 1T+ processes executed per year across the Tray.ai platform, on one contract rather than a monthly credit allowance. Tray.ai’s published limits are per step, not per run: 45 seconds by default for a connector call and 15 minutes for an HTTP response, and a parent workflow waiting on a callable workflow has no timeout. A long backfill does not have to be cut into 40-minute pieces.
- Agents and MCP with governance built in. Make has agents and an MCP server too, so the difference is what sits around them. Merlin Agent Builder uses Agent Hub to coordinate sub-agents that share data sources and tools. Agent Gateway for MCP publishes your own composite MCP tools from 700+ connectors, with access control, rate limiting and audit logs. Make AI Agents are still labelled beta on Make’s own pricing page.
- Coding assistants as a way to build. Tray Headless ships plugins for Claude Code and Codex, plus Headless MCP for other assistants, so developers build and change workflows from the tool they already use. Make’s equivalent is Maia inside the Make editor and scenario editing through its MCP server.
- Enterprise governance that holds up. SOC 1 and SOC 2 Type 2 audited, HIPAA and GDPR, with audit trails on every plan. Make’s security page, read 2026-10-08, lists SOC 2 Type 2, SOC 3, GDPR and ISO 27001 for Enterprise, and does not mention HIPAA. If you handle health data, ask both vendors for their position in writing: this is the one area where you should trust a signed document and nothing else.
- Complex orchestration. Branching logic, error handling, retry policies and multi-step transformations across many systems.
- One platform under one contract. Cue cut custom integration builds from about a month to about two days. Bynder went from 2–3 integration releases a quarter to 7–8. HackerOne reports 4× developer velocity against its own previous custom builds. Each of those is a published customer story, linked so you can read the detail rather than take the number on trust.
Pricing reality
Make prices in credits, and publishes the rates: Free, Core at $12 a month, Pro at $21, Teams at $38, each for 10,000 credits a month, with Enterprise quoted. Checked against Make’s published pricing page on 8 October 2026. Make’s FAQ dates the change: “As of August 27, credits are the billing unit in Make.”
Most actions consume one credit. Some advanced features that use Make’s AI provider may use more, so an AI-heavy scenario can cost more per run than the same scenario without the AI steps. At small volume the numbers look attractive. At enterprise volume, with high-frequency triggers and multi-step scenarios running thousands of times a day, the bill grows in ways the initial estimate doesn’t predict. Ask for a projection at 10× your current estimated volume, with the AI steps included, before signing.
Tray.ai is enterprise / quote-based. One contract covers the platform, and the bill doesn’t move with trigger volume.
Whichever way you go, price the workload you expect in year two, not the one you are piloting.