Launching an AI Agent for Microsoft 365

The problem

Every workplace software vendor was adding “AI” to their product, and most of it was the same thing: a chatbot that sat idle until someone thought to ask it a question. LiveTiles had built something different, where employees could chat with the company’s own knowledge base and get the exact information they needed in the moment, across SharePoint, Teams, and other workplace apps.

But a strong product does not sell itself in a crowded market, and the instinct inside the company was to market the features, list everything the AI could do, and trust buyers to notice. That instinct would have dropped us right back in the chatbot pile, because a buyer who can’t tell what something is never gets as far as deciding whether it’s good.

The solution

Positioning had to come before messaging, so I started with the market and the buyer. I researched how analysts framed this space, including the Gartner Magic Quadrant for packaged intranet solutions, and how IT leaders and knowledge workers used AI day to day.

The research pointed to a precise fit: what we had built behaved like an agent, acting on the employee’s behalf to surface the right knowledge, not an assistant waiting to be prompted. “AI Agent” was the language the market and the analysts were already moving toward, so that was the position, and it had to hold everywhere.

Holding it was the hard part, and harder because this was a new product in the LiveTiles line, not an update to something teams already knew how to sell. The move most of the company wanted was to lead with features, and feature marketing would have buried the one thing that actually separated us: what the product was, not the length of its capability list.

So I kept the positioning fixed and put heavy enablement behind it, training EMEA, APAC, and the US to sell a product none of them had sold before and to tell one story instead of three regional versions, leading with why the product was different rather than what it could do. By the time a prospect reached a rep, they already understood the category, so the conversation was about fit, not explanation.

The whole go-to-market ran six months, from the first research through a coordinated launch across three regions.

The outcomes

Industry analysts called the product “innovative and groundbreaking,” singling out the agent approach as what set it apart.

That carried more weight than any line we could write about ourselves, because third-party validation told buyers what the product was in words sales could never credibly claim on its own.

Adoption came in 2-3x higher than previous LiveTiles launches, retention held as the product kept proving useful in daily work, and the six-month model became the template later launches started from instead of beginning again.