2025 Year in Review - Nebuly - Enterprise user analytics for AI agents

2025 Year in Review

TLDR

→ 2025 was the year enterprise AI moved from experimentation to production. Nebuly's customers stopped asking "will this work?" and started asking "is this working, and what is it worth?" → Nebuly reached production maturity in 2025, with customers across sectors and countries using it to understand AI adoption, measure productivity gains, and track ROI at the organizational level. → The common thread across every customer deployment: the need to understand how AI is actually used, not just whether systems are running. → 2026 brings more complex deployments, more autonomous AI agents, and higher expectations for demonstrable business value. The measurement layer becomes more important, not less, as AI scales.

2025 was a meaningful year for Nebuly, and I wanted to share a few reflections before we head into 2026.

Over the past year, our product reached a new level of maturity. Customers now tell us it covers most of the real challenges that appear once GenAI moves into production. Understanding users. Measuring value. Tracking ROI. Operating with confidence.

We are now live in production with world-class companies across sectors and countries. Each use case is different. What they all have in common is the need to understand how AI is actually used and whether it delivers real value.

Several beliefs we had early on were confirmed in 2025:

AI is not becoming a monopolistic market. Enterprises rely on many models and many GenAI products at the same time. This makes a centralized view of AI usage and impact essential.
From 2024 to 2025, teams moved from experimentation to production. The focus changed. Less exploration. More responsibility. GenAI now needs to work reliably, at scale, with a clear business case behind it.

This is where Nebuly increasingly sits in the stack. When AI becomes part of daily operations, visibility into user behavior and outcomes is no longer optional.

2025 was also the year we strengthened how we work with others.

We started collaborating closely with leading partners such as Reply, Accenture, IBM, and FT Strategies. Large organizations need more than a product. They need support, integration, and governance. These partnerships help us meet customers where they are.

Looking ahead to 2026, AI adoption will keep accelerating. Most large companies will run GenAI in production across at least one team.

Internal use cases will continue to lead the way. Employee-facing copilots remain the fastest path to value. But more teams will then move toward customer-facing use cases. The upside is higher. Expectations are higher, too. User experience, trust, and safety will matter more than ever.

Companies will need to understand how people interact with AI. What works. What does not. Where value is created. Guessing will not scale.

For us, 2026 is about staying focused. Listening closely to customers. Continuing to build with real usage in mind.

Thank you for being with us this year. To our customers, partners, and everyone following our work, we truly appreciate the trust and the conversations along the way.

Looking forward to what we will build together next year.

Julien,

Co-founder

(Book a 1:1 with me)

New Posts

Jul 8, 2026

Insights

Build vs buy: what it takes to build AI analytics in-house

Enterprises tell us an in-house AI analytics build prices north of $1M and 12 to 18 months. A decision guide for enterprise AI leaders.

Mar 18, 2026

Product

Your customers already tell your AI what they want. Now you can see it.

A customer asks your insurance AI about adding travel cover. The AI answers the question and moves on. But that question contained a purchase signal - the customer just told you what they want to buy next. Upsell signals appear in customer AI conversations every day, across every industry. This article explains what they look like, how they are captured at scale, and what commercial, product, and CX teams can do with them.

Mar 16, 2026

Product

AI Proficiency: the metric that tells you whether employees are getting real value from AI

McKinsey found a 6x engagement gap between AI power users and typical employees in the same organizations. EY found that 88% of employees use AI daily but only 5% use it in advanced ways. High adoption and low proficiency is the most common state enterprise AI deployments find themselves in. AI Proficiency is the metric that measures the gap: not whether employees are using AI tools, but how effectively those interactions are actually working. This article explains what AI Proficiency measures, how it is calculated from behavioral signals in conversation data, and why it is the leading indicator of whether your AI investment will generate lasting business value.