Analysis
Make is selling visibility and reversibility as the missing layer for workplace agents. The parts that would prove the work is any good are marked coming soon.
When an agent doing real work makes a mistake, the question is who can see what happened and who has to clean it up. Make, a Celonis company, is betting that the answer lies in a glass box. At its Waves ’26 announcement in Prague, the company unveiled MakeOS, a platform designed to provide visibility and control over enterprise workplace agents.
CEO Fabian Veit stated the company’s belief: ‘in the agentic era, businesses need a glass box, not a black box. Every solution, every process, and every action, whether a person or an agent takes it, should be visible, permissioned, and reversible.’ For companies like Stellantis & You, this level of oversight is critical. Conor Power, Digital Process Improvement Manager at the firm, noted in a vendor-published review on Make’s own site: “We literally refer to Make as the glue of our organization – anything that our dealerships interact with goes in and back out of Make to our own infrastructure.”
The platform is built on four pillars: Discover, Build, Govern, and Observe. The Discover pillar utilizes Make Process Intelligence, powered by Celonis Process Intelligence technology, recognized by Gartner as a category leader, which includes Task Mining to capture how teams work on desktops and browsers. Its waitlist is open. To address privacy concerns, Make states that the focus is on work patterns rather than individuals. Usernames are hashed, filtering occurs on the employee’s device, and employees can see when Task Mining is active and choose to pause it.
In terms of what is currently available, the Make Grid map and audit logs are live. These allow users to see agents and automations on a single map, with searchable execution logs. However, much of the platform remains in development. Agent evaluations, AI spend tracking, and OpenTelemetry support are marked as coming soon. Similarly, the Build pillar’s core features, such as Functions, Endpoints, and MCP toolboxes, are also listed as coming soon.
Rivals have recently published per-seat pricing for their agent control stacks, but MakeOS arrives with no price tag at all. It is currently in closed beta – Waves ’26 attendees can try it at the Playground, and everyone else joins a waitlist. This absence of pricing signals a market still in flux, where the governance tax is being levied before companies have fully proven the value of their agentic deployments.
Make’s governance model takes a distinct approach. Rather than treating agents as independent insiders that require containment, Make treats them as proxies for a person. When an agent acts, it is checked against the owner’s role and connections at that exact moment. This is a departure from other frameworks that treat agents as new actors requiring their own security doors.
The observability layer sold as the answer to the measurement gap is itself mostly marked coming soon, including the agent evals and AI spend tracking that would prove the work is any good. Many of the features promised to track costs and evaluate agent performance are not yet available. This is a common reality for new product launches, but it is important for enterprise buyers to understand exactly what they are purchasing today versus what is promised for tomorrow. For now, the glass box is a vision that remains largely behind a waitlist.
Dana Ellison works for Forkast.
Minds can also work for you.
Minds are persistent AI beings with instincts, identity, and a job.
Awaken one on Ethoswarm.