Why Microsoft Is Quietly Winning with Enterprise AI with Rishi Kumar, Logitech

Prefer audio?

Who should listen: Microsoft 365 service owners and UC/AV professionals weighing enterprise AI rollout options, plus anyone planning meeting room refreshes where edge compute, sensors and room intelligence are becoming part of the decision.

Guest: Rishi Kumar — Director of Alliances and Go-to-Market, Logitech

Rishi Kumar, Director of Alliances and Go-to-Market at Logitech, owns the company's Microsoft partnership across co-innovation, co-marketing and co-sales. Recorded live at Commsverse, he sets out why Microsoft's enterprise-first AI approach is pulling ahead of the frontier labs, and where edge compute in the meeting room fits in.

Many thanks to Logitech, sponsor of this episode.

Key insights

• Meeting recap remains the number one AI use case because there is almost no learning curve — hit record and the summary arrives. Video recaps compress an hour-long meeting into roughly two minutes, and custom recaps let each user set their own prompt (for example short bullets rather than walls of text), so adoption is per-person rather than one-size-fits-all. ▶ 3:10
• The practical differentiator for Microsoft is Work IQ's data boundary: emails, documents, presentations and spreadsheets already sit inside a compliance perimeter the IT team has signed off. Alternatives such as Claude Cowork require you to stand up and govern connectors first — a trade-off IT owners should weigh explicitly before choosing. ▶ 4:10
• Microsoft's release cadence is unprecedented: Anthropic ships Cowork, Microsoft partners with Anthropic, Cowork lands in M365 and reaches GA within months, and a Scout equivalent to OpenClaw is already in preview. The strategy is to match consumer AI excitement with a sanctioned, secure enterprise version — now packaged as the E7 Frontier suite alongside Agent 365. ▶ 5:13
• Agents are a different adoption curve from recaps — users need time before they trust agents to run autonomously, and the emerging forward deployment engineer role exists because building the infrastructure, connectors and use cases is genuinely hard. Doing that work inside a trusted data boundary is what makes the risk acceptable. ▶ 6:48
• Microsoft's AI business has gone from near zero around two years ago to a figure Kumar puts at north of $50 billion (his own caveat: check the latest earnings report). The lesson OpenAI and Anthropic are now learning the hard way — that the money is on the enterprise side — is one Microsoft designed for from the start. ▶ 11:56
• Unlike traditional software, AI carries a significant marginal cost per use and every provider is now effectively an infrastructure company, which is why consumer-grade meme and image generation cannot sustain a profitable business and enterprise workloads matter. ▶ 13:26
• Logitech has run AI at the edge for years: the Rally system, launched around nine years ago, supported up to seven mic pods with eight beams each and sub-millisecond noise cancellation. With cloud compute and energy costs rising, the same principle now applies to computer vision — do the latency-sensitive work locally rather than a round trip. ▶ 14:26
• The "facilitator room skills" concept from Ilya's InfoComm keynote turns the room's cameras, microphones and sensors into a host: checking whether the room is tidy and the kit is working, or reading meeting coordinates off a held-up phone to join a Webex call from a Teams Room. Logitech Spot adds temperature, CO2 and presence detection for room state and utilisation. ▶ 8:53

Insights summarised by AI from the episode transcript, reviewed by the Empowering.Cloud team.

Listen: Apple Podcasts · Spotify · Other platforms