Scaling Sigma Agents Beyond the Workbook: Distribution, Management & What's Next

Today marks the last installment in our "Month of Agents" series. Earlier this month we introduced the new Agent Builder Assistant ("ABA") for building Sigma Agents, covered how to govern and permission them, and demonstrated how to power them with data models, tools, and observability. This week, our team looks at what's next: agents that run outside the workbook, open distribution to any interface, and admin controls to manage everything at scale.

In this episode, we cover:

• Running Sigma Agents outside the workbook, so they can work on a schedule or operate independently of any single workbook
• Accessing and invoking agents from any platform with an API and MCP
• Managing agents at scale with new admin visibility
• Evaluating agent performance over time

Watch the video above to see it all in action, or read the accompanying transcript to understand how our Product and Engineering teams are approaching the future of Sigma Agents.

As a public beta feature, Sigma Agents are accessible by default to any Sigma customer with an active AI provider enabled. Reach out to your Sigma representative with any questions. If you're not a Sigma customer yet, you can request a demo or start a free trial to get started.

Read on to see the full video transcript.

What's next for Sigma Agents?

Zalak: Hello and welcome everyone to episode four of our agent series. My name is Zalak Trivedi, and I'm a Director of Product at Sigma. Over the last few weeks, we've learned a lot about agents, all the things we have today in our product since the launch of agents. Today it's a little different: we're going to talk about the future, and where agents are going.

With me, I have Chris and Jisha on this episode. Chris, agents have been super vital, we have a lot of customers using them. Over the last few episodes, we've talked about a lot of what we have in agents as a product. Where are we going next? What are you cooking?

Why are agents moving outside the workbook?

Chris: Right now, agents live in a workbook, and they help you use that workbook and everything that comes with it. We're trying to evolve into more of an agent platform, and a big step in that is bringing agents outside the workbook. This is going to be great because it's much more flexible and adapts to users' needs. Agents can be more personalized, they can run on a schedule, gather up the information you need, and do work on your behalf, or your business's behalf, independent of any workbook in Sigma. We think this is going to be a great evolution in how our customers use agents.

Zalak: Very cool. This is clearly a mental model shift. Before, things were locked inside the workbook. Talk to me about what this change means for our customers.

Chris: It means a lot more flexibility in how you go about certain things. When you're confined to a workbook, there's a lot you need to know about how workbooks work, and that workbook may have other users and other things to keep in mind. Whereas outside the workbook, you can do interesting things, like talk to multiple workbooks at the same time, or just do work for yourself, on your own behalf, or for a particular team. It's about being agent-first, as opposed to workbook-first. That's going to be exciting.

How will agents be distributed inside and outside Sigma?

Zalak: We're making building agents really easy outside the workbook. Jisha, can you talk about distribution, both inside and outside Sigma?

Jisha: We want to make distribution of agents really easy too, and we're not opinionated about which interface you're using agents from. We'll have an API and an MCP for agents, so you can use them in Claude, Codex, Coco, wherever you get work done. You can invoke agents and take action with agents from just about anywhere.

Zalak: I really like this philosophy of Sigma not being opinionated about the interface of interaction. It could be Claude, Codex, Coco, anything you mentioned, and we just become the engine powering that interaction. That makes a lot of sense.

How do you manage agent sprawl at scale?

Zalak: While you're distributing these agents everywhere, what about management? How do you plan on managing the large scale of agents people are creating?

Jisha: With agents proliferating across your ecosystem, management in the agents platform is really key, and a key focus for us. Soon, as an admin, I'll be able to see all the agents created in my organization, and see usage and token usage models for those agents. If Chris is blowing up usage on one of those agents, I can take action on it, and I can set roles and permissioning controls. We'll have full transparency and management controls for admins, across all their agents, on the Sigma platform.

Zalak: This is very much focused on the IT persona, someone who wants people to leverage the best of AI tools, but also have full transparency and control over the platform people are developing on.

Why does Sigma provide evaluations for agents?

Zalak: To make sure these agents stay consistent over time, I know we run evals internally on our own AI products. What about providing evals for the agents customers are creating? Chris, want to talk about that?

Chris: One thing that's really valuable in an agent platform is understanding not just whether your agents are being used, but how effective they are. That's the step-function change that needs to happen, and we're providing that data to customers: here are all the different questions being asked of your agent. But how do you take that and actually make your agent better? That's what we're going to invest in, and that's the platform we're going to provide for our customers soon.

Zalak: So for a Sigma customer, you should be able to build, distribute, govern, and manage things at scale, using Sigma as an agent platform.

Chris: Yeah, that's exactly right.

When will this be available?

Zalak: Very cool, that was a lot of great things. Thank you all for watching the agent series. But before we wrap, one last question: when will all of this be available to our customers?

Jisha: End of Q3, so very soon.

Zalak: Awesome. Thanks a lot, guys, for joining this episode. And that's a wrap.