I’m proud to announce the newest release of ModelOp’s Enterprise AI Command Center: ModelOp Center 3.5.
ModelOp Center 3.5 takes deeper control of AI agent sprawl. It unifies oversight of low-code, no-code and pro-code agents across the platforms where enterprises are building agents fastest: Microsoft Agent 365 (including Copilot Studio and Foundry), Amazon Bedrock AgentCore and Amazon Quick. Discovered agents and shadow AI flow straight into automated intake and approval, and a new AI usage dashboard expands AI FinOps with use case-level token tracking.
ModelOp Center 3.5 at a glance
- Deeper, out-of-the-box integrations with Microsoft Agent 365 (including Copilot Studio and Foundry), Amazon Bedrock AgentCore, Amazon Bedrock and Amazon Quick, alongside OpenAI, MCP, and more
- Discovery that goes beyond the inventory: each newly found agent moves from unmanaged to governed through risk-proportional workflows
- Expanded AI FinOps, with an AI usage dashboard that ties token usage to the business use case, model and provider
- Lifecycle stage duration tracking to find delivery bottlenecks, plus simpler navigation
Built With the Fortune 500, for Every Team That Delivers AI
Every capability in 3.5 started as a conversation with a customer. Our customers include Fortune 500 banks, insurers, regulators, healthcare and public sector organizations, manufacturers, and retailers running AI on more platforms every quarter. We build with them, test at their scale and under their regulators. What you see in 3.5 is what matters to teams industrializing AI delivery and governance at the enterprise level.
That's also why ModelOp is built the way it is. Delivering AI in a large enterprise takes business line owners, model and solution owners, governance leaders, and control partners in legal, risk and compliance, all working from the same record. ModelOp isn't a data science platform built for data scientists, and it isn't a GRC tool built only for compliance teams. It's the Enterprise AI Command Center where those teams meet: one system of record, shared workflows with enforceable governance, and operational intelligence across ML, GenAI, agentic AI and vendor AI. ITSM tools like ServiceNow handle tickets and assets well, but AI needs a record that can hold one use case backed by several models and agents, each with its own owner, controls and changing risk.
Deeper Microsoft and Amazon Integrations That Go Beyond Discovery
Teams build agents wherever they have access. While that's good for speed, it's how agent sprawl starts. 3.5 deepens ModelOp's out-of-the-box integrations with the platforms where enterprises build and run AI: business-user agents in Microsoft Agent 365, Copilot Studio, and Amazon Quick, developer-built agents in Microsoft Foundry and Amazon Bedrock AgentCore, frontier models from Amazon Bedrock, OpenAI, MCP tools, and more. Discovery runs continuously in the background or as an on-demand audit, and flags anything running that isn't in the approved inventory.
When an agent is imported, ModelOp pulls in the details the platform already holds and keeps the record in sync as the agent changes. Owners don't retype what the platform knows, and the record doesn't drift.
Who it's for: Solution owners get their agents registered without the manual paperwork. Governance leaders and control partners see new agents as they appear, including the ones nobody registered. CIOs get full visibility across Microsoft and Amazon.
From Discovery to Approval, Automatically
A discovered agent now starts a workflow automatically, and that workflow enforces the enterprise's own policies and standards. It's built so agent owners, governance leaders and control partners can all trust it to get AI to production rapidly, safely and profitably. ModelOp registers the agent through a workflow built for its platform and flags it as unmanaged until intake is complete. The owner is asked only for what the platform can't supply, such as intended use, expected value and data sensitivity, and every intake question is configurable.
From there, the workflow handles the reviews, tests, and approvals:
- Approved components. Are the LLM and MCP tools behind the agent approved for production? If not, the owner can request approval in the same flow.
- Test data. If no test data is attached, the owner is prompted to add it, so validation is traceable.
- Right-sized review. An agent matching a pre-approved pattern, such as internal use with no sensitive data, gets a lighter review. One that touches confidential data goes to responsible AI, legal and model risk, in parallel or in sequence.
Who it's for: Agent and solution owners get a guided path to production. Governance leaders get risk-proportional controls applied the same way every time. Control partners in legal, model risk and responsible AI receive complete submissions with the evidence attached.
AI FinOps: From the Token to the Business Use Case
Once an agent is in production, the next questions leaders need to answer are "what does it consume?" and "is it worth it?" 3.5 moves AI cost accountability from the token to the business use case. A new AI usage dashboard in the Enterprise AI Command Center shows token usage by model and, most importantly, by business use case. Because each agent and model is already tied to a use case and owner in the system of record, usage is attributed automatically.
Leaders can see which use cases consume the most, compare similar use cases across models and providers, and decide whether to keep a use case, change its model or retire it. They can drill into a single use case, follow how its usage trends month over month, and get alerts on material increases and new use cases or models.
Who it's for: CIOs, CTOs and business line owners can manage AI as a portfolio, weighing consumption against expected value. Solution owners can compare models for the same job.
Pinpoint Where AI Delivery Stalls
A new lifecycle progress bar and stage duration tracker show how long each use case spends in each stage. For a business owner or CIO measuring AI delivery speed, that turns "why is this taking so long?" into a specific stage, team and fix. Configurable stages and conditional intake forms let each enterprise fit the lifecycle to its own operating model.
We also simplified the navigation. The main menu used to label the inventory "Model," a name from the platform's early days. But ModelOp has always supported every type of AI enterprises run: ML, GenAI, agents and vendor AI. Since we helped pioneer AI governance, we've brought each new type into the platform as it arrived. The menu now says Inventory, and Model Lifecycles is now simply Lifecycles.
Who it's for: CIOs and AI CoE leaders tracking delivery speed, governance leaders tightening their process, and everyone who works in ModelOp day to day.
Built for the Heterogeneous AI Enterprise
Microsoft Agent 365 and Amazon Bedrock AgentCore give teams strong technical control over the agents on each platform. But, neither can align agents with enterprise processes and policies across every platform an enterprise runs. That's the job of the Enterprise AI Command Center: teams keep freedom of choice, while the enterprise has operational control of their vendor agents, models, and tools.
ModelOp was ranked highest for the AI Agent Governance Use Case in the 2026 Gartner® Critical Capabilities for AI Governance Platforms report. We believe 3.5 takes that work deeper into the agent platforms our customers are adopting now, built with and for the enterprises that can't afford to get AI wrong.
Request a demo to see how ModelOp Center 3.5 takes control of agent sprawl, accelerates agent deployment, and brings greater visibility to AI FinOps.
Frequently asked questions
How does ModelOp govern agents built in Microsoft Agent 365, Amazon Bedrock AgentCore, and Amazon Quick?
ModelOp Center 3.5 discovers agents on Microsoft Agent 365 (including CoPilot Studio and Foundry), Amazon Bedrock AgentCore, and Amazon Quick, along with frontier models from Amazon Bedrock, OpenAI, and MCP tools through out-of-the-box integrations and compares them against the approved inventory. Based on policy, each new agent found can automatically start a governance workflow: intake, an initial risk tier, risk-proportional controls and approval routing, with the agent tied to its business use case and owner.
Does ModelOp replace Microsoft Agent 365 or Amazon Bedrock AgentCore or Amazon Quick?
No. Those platforms build, run and secure agents. ModelOp's Enterprise AI Command Center sits above them as the system of record, connecting each agent to its business use case, owner, approvals, controls and usage.
Which platforms does ModelOp Center 3.5 discover AI in?
Microsoft Agent 365 (including Copilot Studio and Foundry), Amazon Bedrock AgentCore, Amazon Bedrock, Amazon Quick, OpenAI, MCP, and more. The platform is built to add integrations as enterprises adopt new AI technology.
Is ModelOp a GRC tool or a data science platform?
Neither. ModelOp is an Enterprise AI Command Center that brings business, technical, governance and control teams into shared workflows above AI development and execution, AI security, GRC, ITSM and data management, without replacing them. ModelOp is an AI delivery platform.
How does AI FinOps in ModelOp track usage by business use case?
Every model and agent is tied to a business use case in ModelOp's system of record, so token usage is attributed to that use case automatically and shown in the AI usage dashboard, with use case-level token tracking across models and providers.
Gartner disclaimer
Gartner, Magic Quadrant for AI Governance Platforms, Lauren Kornutick, Sumit Agarwal, Priya Sundararaman, Nader Henein, Brandon Medford, June 2026.
Gartner Report, Critical Capabilities for AI Governance Platforms, By Sumit Agarwal, Lauren Kornutick, etc., June 2026.
Gartner is the trademark of Gartner, Inc. and/or its affiliates.
Gartner does not endorse any vendor, product or service depicted in its research publications and does not advise technology users to select only those vendors with the highest ratings or other designation. Gartner research publications consist of the opinions of Gartner's research organization and should not be construed as statements of fact. Gartner disclaims all warranties, expressed or implied, with respect to this research, including any warranties of merchantability or fitness for a particular purpose.
