3.0 is not a feature release. It changed where agentic AI is heading.
Three things are different now. Where agentic AI is going, who builds it, and where you start.
The destination changed. There is no agentic fleet to manage.
Everyone else is adding agents. Gartner expects the average large enterprise to be running more than 150,000 of them by 2028, up from fewer than 15 in 2025, and only 13% of companies say they are confident they can govern the ones they already have. (Gartner, April 2026)
We went the other way. You do not deploy agents anymore. You run one intelligent agent that carries your full library of business skills and decides at runtime how each case should be handled. Everything built so far becomes capability inside that one agent.
Contained, not assembled.
The builder changed. The business holds the pen.
Once your technical team sets up the connections, the people closest to the process create, test, and refine agents by talking to MyWave Create™ in plain English.
Customers have built working agents in a single hands-on workshop session.
The people who know the work now build the agents that do it.
The starting line moved. Nobody starts from a blank page.
MyWave builds agent capability from 27 industry ontologies: encoded knowledge of how each industry actually runs its processes. Generation starts at over 80% complete. The business refines the last mile in hours.
What 3.0 delivers
One agent, many skills
A skill is one job the agent knows how to do. Capability grows by adding skills to one library, so governance stays simple as your footprint compounds.
The Supervisor
Every response is checked against your policy before your users see it. The check is deterministic and runs outside the model. Gartner named this category in February 2026 and calls it a guardian agent. (Gartner, February 2026) Ours is the deterministic one.
Plain-language policies
What the agent may do, the directive it works to, the exact tools it can touch. Written and changed in natural language, by the business that owns the process.
Deterministic workflows
Where a process must run identically every time, structured flows execute step by step alongside the autonomous agent. AI builds the flow. Determinism runs it.
Choose the mode per process.
Probabilistic where the work is full of exceptions. Deterministic where compliance depends on it. All four of Gartner's autonomy tiers, set process by process.
Open standards, certified core.
MCP, A2A, and webhooks, every connection through one policy gate. Plus a certified SAP BTP extension (SAP BTP-EXT-GenAI) executing governed transactions in the SAP core.
Probabilistic where it thinks. Governed in everything it does.
What it means for you.
Customers: you were going to run a program. Now you run a process.
Your first agent runs in test within weeks, and your own team is succeeding in the first week. A policy change is one edit, in one plain-language document, reviewed in one cycle.
Global systems integrators: a bespoke build becomes a repeatable pattern you own.
Generation kills the blank-page phase. Your consultants coach and harden instead of hand building, and everything built for client one becomes reusable IP for client two.
SAP Business One partners: sell agentic AI without hiring an AI team.
Three-click deployment on MyWave Cloud™. No BTP: it runs where B1 runs. Build it in the meeting, deploy it in three clicks.