Key takeaways
- Two things became practical at once: observing real work at scale, and AI that can reason about it.
- Data on how people work can feed memory tools, answer tools, automation, surveillance, or skill-building. Only the last leaves people more able.
- Seeing work got cheap. Trust and coaching did not, and that is where the durable advantage lives.
Two conditions crossed together
Two conditions crossed the threshold at the same time: observation of real work at workforce scale, and AI capable of reasoning about that work in real time. Either alone was not enough. Observation without reasoning produced recordings nobody watched. Reasoning without observation produced generic advice. Together they make a new kind of product possible, and the market has noticed: Workday acquired Sana (about $1.1 billion, 2025), and Meta acquired Limitless (2025).
What observation becomes
| Observation becomes | Effect on people |
|---|---|
| Memory | A searchable past. Doesn’t change how you work. |
| Answers | Help in the moment. Dependence over time. |
| Automation | Routes around people instead of building them. |
| Surveillance | Breaks the trust the data depends on. |
| Capability | People who build their own solutions, on an architecture they trust. |
What each path does to people
Automation and process mining route around people, which captures some efficiency but leaves the redesign capability in the vendor’s hands. Surveillance fails on its own terms: the moment people feel watched, they hide their real work, and the data degrades. Capability is the only path that leaves people more able than it found them.
Everyone is converging on the same space
AI model companies are expanding from chat assistants; learning platforms from search and HR; process-mining vendors from workflow maps; desktop-capture tools from recording. No one owns this space yet. Our advantage is a head start in earning trust, not a barrier others can’t cross, so moving early matters more to us than owning ground.
Where the advantage lives
Observation itself is commoditizing. Earned permission and coaching are not. The right to observe real work is granted person by person; it cannot be shipped in a release. That is why we treat trust as architecture and coaching as the product.
If you are evaluating tools in this space, ask three questions: what does the product do with what it sees; who does it work for; and what can people do afterwards that they could not do before? To see how this advantage compounds, read about the three loops.
Sources and notes
- Workday, “Workday Completes Acquisition of Sana,” November 4, 2025; TechCrunch, “Meta acquires AI device startup Limitless,” December 5, 2025.
Founder & CEO of Lucy Labs. Twenty years running enterprise transformation by hand in big tech and Global 2000 companies, now building it into software.
