A practical field guide from Automation Ace.
Short answer: MakeOS is Make's platform for running work with AI agents under visibility and control. Unveiled at Waves '26 in Prague, it builds on Make's “glass box” idea: every agent, process, and action should be visible, permissioned, and reversible. Agents act as proxies for a person, are checked against that person's role when they act, and are blocked and escalated when they exceed their limits. At the time of writing MakeOS was in closed beta with a waitlist, with several capabilities marked as coming soon.
Short answer
- What: an enterprise platform for governed, visible agent work.
- Principle: glass box, not black box.
- Status: closed beta; pricing not announced; some features coming soon.
The glass box idea
As AI agents take on real work, the main risk is opacity: nobody can see what an agent did, why, or with whose permissions. Make's answer is a glass box: show every agent and automation on one map, log every action, check permissions at the moment of action, and make it possible to reverse what went wrong. It is a governance stance as much as a product.
What MakeOS includes
| Idea or capability | What Make has described | Status at the time of writing |
|---|---|---|
| Glass box | Every agent, process, and action visible, permissioned, and reversible | The guiding principle |
| Agents as proxies | Agents act for a person and are checked against the owner's role and connections when they act | Announced |
| Governance | An agent that exceeds its limits is blocked, asks a person, and the attempt is logged | Announced |
| One map | Agents and automations on one map (Make Grid) with searchable execution logs | Announced |
| Sharing | Agents shared across departments without sharing the underlying data | Announced |
| Evaluations, AI spend tracking, OpenTelemetry | Quality checks, cost visibility, streaming to monitoring tools | Listed as coming soon |
| Access | Closed beta with a waitlist; pricing not announced | Closed beta |
Details are based on Make's announcements and coverage at the time of writing and may change; check Make for current scope.
How it fits with the rest of Make
| Make capability | Role in the bigger picture |
|---|---|
| Scenarios | Deterministic workflows and tools for agents |
| Make AI Agents | Agents that use modules, scenarios, and MCP tools |
| Maia | An AI co-worker that builds and fixes scenarios and agents |
| MCP Server and Client | Connecting outside AI clients and tools |
| Make Grid | A live map of automations and agents |
| Process Intelligence | Finding what to automate next with process and task mining |
| Human-in-the-loop approvals | People sign off before risky steps |
Guides for each piece: building AI agents, using Maia, connecting agents with MCP, agent permissions, monitoring agents, and Process Intelligence.
Why it matters for businesses
- Agents are moving from demos to daily work, and governance is the blocker for most companies.
- Visibility builds trust with security, compliance, and leadership.
- Permission checks at action time reduce the risk of agents doing more than their owner could.
- One map helps teams manage sprawl as agents multiply across departments and platforms.
The same pressures are shaping other platforms; see Zapier's AI workspace and why AI agents need automation engineers.
Questions to ask before adopting
- Which capabilities are available now versus planned?
- How are agent permissions defined, enforced, and audited?
- How does it handle agents built on other platforms?
- What does reversing an agent action actually undo?
- How are AI costs tracked and attributed?
- What does it cost, and on which plans?
How to prepare now
- Inventory your scenarios and agents; see auditing Make scenarios.
- Clean up connections and ownership.
- Define permission tiers for agent actions.
- Start logging agent runs and costs.
- Pilot one well-scoped agent with approvals; see combining agents with rules.
New to Make? Create a Make account.
Frequently asked questions
What is MakeOS?
MakeOS is Make’s platform for running work with AI agents under visibility and control, unveiled at Waves ’26. It follows a glass-box approach in which agents, processes, and actions are visible, permissioned, and reversible.
Is MakeOS available?
At the time of writing MakeOS was in closed beta with a waitlist, pricing had not been announced, and capabilities such as agent evaluations, AI spend tracking, and OpenTelemetry support were listed as coming soon.
What does glass box mean for AI agents?
It means the opposite of a black box: every agent and automation is visible on one map, actions are logged and searchable, permissions are checked when agents act, and mistakes can be reversed.
How does MakeOS handle agent permissions?
Make described agents as proxies for a person, checked against the owner’s role and connections at the moment they act. An agent that exceeds its limits is blocked, asks a person, and the attempt is logged.
How should I prepare for MakeOS?
Inventory scenarios and agents, clean up connections and ownership, define permission tiers for agent actions, start logging agent runs and costs, and pilot a well-scoped agent with approvals.
Disclaimer: Make features, module names, plans, and credit usage change over time, and several capabilities announced at Waves '26 (such as MakeOS and Process Intelligence) were in beta or listed as coming soon at the time of writing; confirm current details in the Make Help Center. This article may include links to apps, products, or services; some links may be affiliate links, which means Automation Ace may earn a commission at no extra cost to you.