Home

Blog

Red Hat and Lenovo on What to Build and What to Buy

AI Agents & Orchestration

Customer Stories

Embedded Integration

Enterprise MCP

Red Hat and Lenovo took unscripted questions at WOW 2026 and landed on the same advice for enterprise AI: partner where it counts, govern tightly, and get moving.

Hannah Peacock

Director, Solutions Marketing

Oct 14, 2026

Red Hat and Lenovo took unscripted questions at WOW Las Vegas 2026 and landed on the same advice: partner for the plumbing, govern it tightly, and get moving.

At WOW: World of Workato in Las Vegas, two customers from the main stage took open questions with no slides and no script: Sherry Gentry-Gasper, Sr. Director, Enterprise Architecture & Delivery Acceleration at Red Hat, and Pooja Sathe, Head of AI Ecosystem Partnerships at Lenovo. Will Corkery, CRO at Workato, moderated.

Lenovo is building an AI assistant, QIRA, that takes actions across the devices it ships to consumers and enterprises. Red Hat’s team runs enterprise architecture and delivery acceleration, including the CI/CD pipelines behind its internal SaaS applications. While very different, their answers still converged.

Buy the plumbing, build what makes you different

Lenovo wanted its assistant to act across the apps people use every day, such as Gmail, Outlook, Slack, and YouTube. Pooja said the team weighed building in-house against partnering, and that Workato stood out for how well it understood the problem Lenovo was trying to solve rather than offering a cookie-cutter solution.

The consumer use case also meant Lenovo couldn’t forecast consumption. Pooja said the Workato account team helped structure a deal around that uncertainty. Lenovo’s own differentiator, she explained, is understanding context, not building connectors. So Lenovo built the infrastructure, including an MCP registry, and Workato helped it build, test, and register connectors at scale.

Red Hat reached the same conclusion from the other direction. Its internal integration platform had grown to about 2,500 Java endpoints, maintained by senior engineers. Sherry said those engineers should be solving business problems, not maintaining integrations. Build versus buy was her toughest internal conversation: Red Hat should build what solves real Red Hat customer problems, not routine tasks like generating a quote.

Least privilege is the test for enterprise MCP

Sherry was candid about what she would do differently on day one: spend less time comparing native SaaS MCPs with Workato. Her team follows least privilege and least access, and in her experience many native MCPs were all-or-nothing. Red Hat has since made an architectural decision that its operational MCPs will come from Workato.

Her example: an agent that can touch four fields on a quote is a different risk from one that can write back to an entire Salesforce instance. Red Hat’s distinguished engineers tested the platform against requirements like token exchange and session handling before the business got access.

Speed comes from fewer dependencies and more builders

Red Hat has about 300 builders, roughly 40% outside IT and engineering. Sherry credited three things:

Repeatable patterns. Everything is templatized, so anything committed stays within guardrails.
Fewer dependencies. Teams are organized to minimize handoffs, with federated architects embedded in each and enterprise architects aligned to value streams. Work that writes PII or financial data gets a higher risk rating and architecture review. Read-only dashboards do not.
Deliberate enablement. Each of 52 product teams nominated someone for learning days and certification competitions, with prizes. Red Hat spent about six months hardening observability, alerting, and logging first.

The cultural shift mattered too. A centralized integration team had become the bottleneck, and Sherry said engineers let go of writing every integration once they saw what the speed meant for the business. Her bar:

“An integration should never be the slowest part of your solution.” — Sherry Gentry-Gasper, Sr. Director, Enterprise Architecture & Delivery Acceleration, Red Hat

“An integration should never be the slowest part of your solution.”

— Sherry Gentry-Gasper, Sr. Director, Enterprise Architecture & Delivery Acceleration, Red Hat

Lenovo is attacking the same problem from the compliance side. Pooja described a “shift left” approach that starts security and responsible AI reviews when a feature is defined, not at the end.

Keep humans where judgment lives

Asked where to draw the line on automation, both speakers pointed to people. Sherry said the workforce is anxious about keeping up, and that forcing everyone to use AI for everything would backfire. Longtime employees hold context that isn’t written down.

“If we lose those people, AI doesn’t have that context.” — Sherry Gentry-Gasper, Sr. Director, Enterprise Architecture & Delivery Acceleration, Red Hat

“If we lose those people, AI doesn’t have that context.”

— Sherry Gentry-Gasper, Sr. Director, Enterprise Architecture & Delivery Acceleration, Red Hat

Pooja added that AI can create more work than it saves when no one understands the output, so teams need judgment about when and how much to rely on it. Red Hat keeps humans in the loop where empathy and reasoning are needed, and tries to place review only where it adds value.

Start now, and expect to get some things wrong

Both closed with the same advice.

“Mistakes are fine. Risks are fine. So I would just say start and experiment quickly.” — Pooja Sathe, Head of AI Ecosystem Partnerships, Lenovo

“Mistakes are fine. Risks are fine. So I would just say start and experiment quickly.”

— Pooja Sathe, Head of AI Ecosystem Partnerships, Lenovo

Sherry said that waiting to perfect the framework and KPIs would have cost six months. Red Hat instead ships a release 1.0, measures adoption, and retires what nobody uses.

Next for Lenovo: a marketplace of connectors, agents, and skills for people using its devices, and a path into enterprise devices. Sherry wants enterprise architecture to be known as an accelerator of delivery, with pre-built, secure MCPs as the architecture runway.

The takeaway

The advice from both customers is practical: partner where you aren’t differentiated, insist on least-privilege access, design teams around fewer dependencies, and protect the human judgment AI can’t replace. Then start.

The future of work is being planned at World of Workato. Save the date for WOW 2027.