Running Remote turned ten years old in Austin, and Liam Martin opened the conference by doing something rare for anniversary events: he didn’t look back. He looked forward, and drew a sharp line between the decade just passed and the one now beginning.
The data on where we stand is clear. Stanford research from Nick Bloom’s Work From Home Research group shows remote work has stabilised at 27% of the US workforce for three consecutive years — 2024, 2025, and into 2026. Not declining, not surging. Stable. Before COVID, that number was 7%. It surged to 62% in March 2020, then settled. The debate about whether remote work is dying is, Martin argued, a last decade’s headline. Remote is a permanent fixture of how knowledge work operates.
But stability in location is not the interesting thing happening. The interesting thing is the question of who does the work.
The shift from distributed people to distributed intelligence
The next five years, Martin proposed, will not be primarily about where people work. They will be about who does the work — and the answer will include humans, AI, and AI agents working alongside each other. The phrase he used to frame the decade ahead: from distributed people to distributed intelligence.
This is not a comfortable shift. It surfaces three deep challenges that the remote work community has been trying to solve for a decade, and which AI now makes impossible to defer.
The identity crisis
For over a hundred years, the manager’s job followed a predictable pattern: gather people, assign tasks, manage completion, identify quality. Remote work disrupted that model. AI, Martin argued, is finishing the job.
Asana data cited in the keynote shows that 60% of a knowledge worker’s day goes to what they call ‘work about work’ — moving information internally, managing status updates, coordinating across tools and teams. AI is absorbing that work rapidly. The manager who defined their value through supervising that output is facing an identity crisis.
The answer, Martin proposed, is not to manage harder but to manage differently: curate alignment, culture, and judgment. Stop supervising output. Start designing the conditions for good decisions.
The trust gap
For the last five years, the trust question in remote work was: are they actually working? The next five years will ask a fundamentally different question: are they actually real?
Martin cited FBI and DOJ raids on over a thousand companies in 2025 following the discovery that remote workers at those organisations were actually North Korean operatives collecting paychecks while extracting information. This is not a niche security concern. It is now a mainstream operational risk with documented, large-scale precedent.
The shift required: from visibility to verifiability. Not ‘can I see them working?’ but ‘can I verify who they are, what they produced, and whether it is trustworthy?’ Remote work at scale requires building verification infrastructure, not surveillance theatre.
The performance paradox
Martin made this point with a piece of self-deprecating precision. His entire conference presentation — slides, structure, content — took 78 seconds to create with AI. Choosing his jacket took three weeks.
In a world where substantial work products are generated in seconds, measuring performance through effort is no longer coherent. The traditional proxy — working hard, visible busyness, long hours — becomes meaningless when AI can do the equivalent of days of production in under a minute.
Martin proposed a new performance equation: decisions × quality ÷ time. The number of decisions made, multiplied by their quality, divided by the time taken to make them. This is what matters when the cost of production collapses. Human value becomes primarily a function of judgment — how quickly and accurately someone can read a situation, make a call, and take responsibility for it.
The shape of what comes next
Running Remote 2026, Martin framed, is about defining the next decade rather than defending the last. The speakers over the following two days would each be exploring a different dimension of the transition from distributed people to distributed intelligence.
The most useful sentence in the keynote was also one of the simplest. Wave one solved location. Wave two redefines performance. The organisations that understand this early, that stop measuring effort and start measuring judgment, that build verification into their hiring and trust into their culture — those are the ones that will lead the next decade of distributed work. Not the ones still debating whether 27% is enough.
