At World of Workato (WOW) 2026, the annual user conference for enterprise orchestration and AI, three Workato Embedded customers compared how they decide what an agent can touch.
Their finding: An AI agent doesn’t have to be malicious to cost you money. In some cases, it only has to offer a discount nobody approved.
In Governing Agentic AI at Scale, moderator Justin Paul, Director, Solutions Consulting at Workato, sat down with:
- Steve Micklewright, Integration Services Principal, LogicGate: a GRC platform for third-party risk, regulatory compliance, and cyber risk.
- Robert Harris, Director of Product Operations, Nextiva: a customer experience and unified communications platform.
- Anand Narasimhan, CTO, S-Docs: document generation and e-signature for public sector and regulated enterprises.
Each company sits at a different point on the agentic curve and brought a different perspective to four foundational AI questions.
Question 1: What can the agent see?
Answer: Ask what your agent inherits, not just what you grant it.
LogicGate already tied permissions to roles for human users. A role decides which applications, data, and workflows someone sees. When Steve’s team built agents, they were assigned roles from the same model, so governing them needed no new build. According to Steve, this process had worked for years for their end users, so why reinvent it now?
LogicGate’s internal agents follow the same logic. They read GitHub, Confluence, and Zendesk’s knowledge base, and they aren’t connected to Salesforce or its HR system, so no customer or employee data is in reach.
At Nextiva, Robert described a narrower version. Its support agent sees only part of Salesforce and returns only what belongs to the customer asking. He applied the same idea to his customers’ agents: a shipping company might let callers track an order but not start a return without talking to a real person first.
“I wonder how many people out there have implementations where they’re probably giving access to too much information and they don’t even notice.” — Robert Harris, Nextiva
Anand pushed the question past the agent. If a connected system has weak role-based access (RBAC), the agent inherits that weakness, however well the orchestration layer is governed.
Question 2: What can it decide?
Answer: The human step works because the agent shows its work.
Robert’s example came from Nextiva’s customers, which include emergency vets. When a caller’s situation is urgent, the agent can’t keep running them through its loop. It has to recognize the moment and hand the call to a person.
LogicGate applies the same rule to vendor onboarding. Reviewing a new vendor used to mean gathering questionnaires and external risk data and passing them through rounds of approval, where people often reviewed the same data again. That could take days, weeks, or months. Now, an agent reviews the questionnaires and vendor analytics. At the last step, it recommends a decision. A person approves.
“We don’t give them an AI summary per se. We show how the AI made the decision on why what they’re doing is a safe decision.” — Steve Micklewright, LogicGate
Anand’s version is contracts. S-Docs generates proposals, NDAs, and MSAs, so he knows what a stray clause can do. An agent that quietly adds a 20% discount for a missed SLA, he said, would send any legal team into a meltdown. S-Docs’ answer is to become a governed agentic platform for enterprise documents: governed, secure, and simple enough for non-technical users.
Question 3: How will you know if the agent is wrong?
Answer: It’s complicated.
Each panelist answered differently, and those differences are useful, but they call for some self-reflection first. Their advice: how loose you are with your agents depends on what a wrong answer costs.
Anand put evals on his short list, next to role-based control at every level and guardrails.
“If you’re doing agents without evals, you’re kind of shooting in the dark. Hope it works.” — Anand Narasimhan, S-Docs
Steve started from the cost of being wrong. In GRC, an answer can’t be a “maybe.” In his words, if AI told an international bank it might be compliant with a regulation, that bank could face a $100 million fine the next day.
Governance was the easy part for LogicGate, but training the AI took time. Once testing produced good-to-great answers, the team released the agent to the largest user base it could, with guardrails in place and no full access. Every user prompts differently, so feedback arrived the next day and shaped the next version.
Robert, on the other hand, added a signal evals can’t capture: how the end customer feels. Nextiva’s customers put agents in front of their own customers, so he tells them to stay involved and watch sentiment, not assume people will love it.
Question 4: How do you do it again, faster?
Answer: Speed comes from what they didn’t have to rebuild.
LogicGate shows what reuse looks like. Employees talk to the internal agent in Slack. Workato connects Slack to AI APIs and memory stores, with recipes that copy knowledge bases into those stores. Workato also queues the jobs, so many people can ask at once, and the team didn’t have to design for scale.
The first agent, for the integration services team, took two to three months. The second shipped within 48 hours because the recipes were ready to go: a message from this Slack channel goes to that memory store, and the API calls stay the same. The following week, the product management team added one line to a recipe and had its own agent the next day.
“As long as [the agent] has the training, the architecture remains the same. It’s just rinse and repeat. Clone a recipe, point it elsewhere.” — Steve Micklewright, LogicGate
Anand took the same view of governance itself. Any team can reinvent it as a control plane, he said, but why, when a platform already provides one?
Build governance you can reuse with Workato
For the companies on this stage, governance wasn’t a hurdle. It was something they could reuse.
Workato ONE for SaaS Platforms gives product teams that foundation to build on – one governed platform for every workflow, agent, and integration your product ships, with verified user access, human-in-the-loop approvals, and audit trails built in.
Explore Workato ONE for SaaS Platforms