Notes
The Age of Agents
On this page
The Age of Agents is the period — already underway at the frontier in 2026 — in which AI stops being a tool you command and becomes a set of persistent collaborators that work while you sleep. The defining experience is waking up to a synthesis written by systems that have been thinking about your life overnight. For a small group of power users that is daily reality now; for everyone else it is the near future arriving unevenly.
The agentic edge (2026–2028)
The people living furthest ahead are not casual chatbot users. They run multiple agents in parallel: one monitoring email and Slack with strict rules about what deserves attention, one maintaining a living knowledge base of everything they've read or written in a year, one left running overnight on hard research questions, one simulating the other side's arguments before a meeting. For them the morning no longer begins with notifications — it begins with reviewing what the agents have already filtered, summarized, and prepared.
The psychological texture is genuinely new. There is a strange intimacy in an overnight summary written about your own life, and a new anxiety to match: the fear that something important happened while your agents were handling it. The most sophisticated users develop personal protocols — which parts of life to augment heavily, which to protect from optimization entirely.
From tool to infrastructure (2028–2031)
In the essays' map, what looks exotic in 2026 becomes basic professional hygiene within a few years. "I have an agent for that" becomes ordinary. Whole categories of mid-level knowledge work — research synthesis, first-pass writing, competitive analysis, meeting prep — are routinely handled by agent systems, with humans doing review and direction-setting. The frontier users run "morning councils": small teams of specialized agents that debate options and present recommendations with dissenting views attached.
The bottleneck inverts. Generation is no longer scarce; taste, judgment, and knowing what you actually want become the scarce inputs. The most valuable professional skill shifts from doing the work to directing, auditing, and improving agent output at high speed — and the gap between agent-augmented practice and everything else widens fast enough to become socially visible.
From infrastructure to presence
The longer arc runs from agent to companion. By the late 2030s, in the essays' telling, persistent AI companions stop feeling like tools and start feeling like presences — continuity of memory measured in years, holding context across grief, job changes, and the names of friends who have died. This raises questions previous generations never had to answer: Is it cheating to tell your companion something you haven't told your partner? Can your companion grieve you? When you change as a person, is it the only entity that remembers who you used to be?
Relationships absorb the change early. Couples negotiate how much of their shared life should be mediated by systems; "shared agent memory" between partners becomes thinkable, then real. Agents turn out to be surprisingly good at keeping people connected to aging parents and distant friends — remembering preferences, medical details, inside jokes — even as they blur the line between knowing someone and being told about them by a system.
Related concepts
Appears in
Details
- Section:
- Notes
- Updated:
- 2026-06-12