In late August 2026, Cisco started giving a personal AI agent to every one of its roughly 90,000 employees — one of the largest workforce-wide agent rollouts any company has attempted. Here is the direct answer to the question everyone’s asking: an AI agent for every employee makes each person faster, but it does almost nothing for how their work adds up as a team — and at 90,000 agents, the coordination problem gets worse, not better. Lova is the fix for that gap: a chat-first AI project management product where AI agents work as teammates, claiming bounded tasks on a shared board, moving them through explicit states, and leaving an audit trail every teammate — human or agent — can read.
Key takeaways
- Cisco is rolling out a personal agent called MyAgent to about 90,000 employees, moving work “beyond chat and into supervised autonomous execution” across tools like Outlook, Webex, Jira, and SharePoint.
- The rollout is a maximal bet on the individual. But Microsoft’s 2026 data finds organizational factors drive 67% of AI’s impact versus 32% for individual skill — so a personal agent invests in the smaller half of the equation.
- Coordination is already the bottleneck. Atlassian’s State of Teams 2026 pegs the fragmentation tax at $161 billion a year on the Fortune 500, with 87% of knowledge workers saying they lack the time or capacity to coordinate.
- Agents left to run without a shared surface drift. An Enterprise Management Associates report found 65% of enterprises have already seen AI agents act outside their intended scope, with 29% reporting tangible organizational consequences.
- Capability isn’t the risk — coordination is. Gartner projects that over 40% of agentic AI projects will be canceled by 2027, largely on governance and unclear value.
What is Cisco’s MyAgent, and why does it matter?
MyAgent is a personal AI agent Cisco is handing to its entire workforce — not a pilot team, not one department, but everyone. It moves, in the company’s own words, “beyond chat and into supervised autonomous execution”: an employee gives it a goal, and it uses approved data and applications — Outlook, Webex, Jira, SharePoint — to push the task forward. It runs on Cisco’s internal platform and, for cost control, routes most requests to cheaper models: reporting on the rollout describes roughly 50–60% of requests going to open-weight models, 20–30% to plain software automation, and only a small remainder to a frontier model. Cisco’s president and chief product officer, Jeetu Patel, framed the stakes plainly: “AI agents reason and act continuously at software speed, and that changes everything about how we scale, manage, and defend” the work.
It matters because it makes a bet visible that most companies are making quietly. The default corporate AI strategy of 2026 has become: give every person their own agent and let a thousand desks get faster. Cisco just did it at the largest scale yet, and did it in the open. That makes it the cleanest test we have of a question the whole market is about to face — if you give everyone a personal agent, what actually happens to the organization?
The context sharpens the question. In May 2026, Cisco told staff it would cut fewer than 4,000 roles — under 5% of its workforce — framing it as a reallocation toward AI, and weeks later began handing the remaining employees a personal agent. Fewer people, an agent each, more autonomous work in flight. The productivity math is obvious. The coordination math is the part nobody has solved.
Why does a personal AI agent for everyone speed up individuals but not the org?
Because a personal agent optimizes the desk, and organizations don’t win or lose at the desk. This is the finding that should reframe the entire rollout. Microsoft’s 2026 Work Trend Index — a global survey of 20,000 people — tested 29 factors behind whether AI actually pays off and found that organizational factors (culture, manager support, how work is structured and governed) account for 67% of AI’s real impact, more than twice the 32% that traces to individual skill and mindset. Training a person, or handing them a capable agent, moves the 32%. It leaves the 67% untouched.
A personal agent is, almost by definition, an investment in that smaller half. It lives on one person’s context, works their inbox, advances their tasks. What it cannot do is see the seams — the handoffs, the dependencies, the place where your finished work becomes someone else’s starting point. And the seams are exactly where 2026’s pain lives. Atlassian’s State of Teams 2026, drawn from more than 12,000 knowledge workers, found the Fortune 500 loses $161 billion a year to coordination friction, that 87% of workers say they lack the time or capacity to coordinate with everyone stuck in execution mode, even as 89% of executives insist AI is making them faster. Faster at the desk, slower between desks. That is the productivity paradox in one sentence, and a personal agent per employee is its purest expression.
The personal-agent paradox
Here is the frame worth naming, because the market doesn’t have a name for it yet. Call it the personal-agent paradox: the more personal the agent, the more fragmented the organization — unless the work lands somewhere shared. A personal agent is bound to one person on purpose; that’s what makes it feel magical and what makes it structurally blind to everyone else. Give one person an agent and you get a faster person. Give 90,000 people an agent each and you don’t get a faster company — you get 90,000 faster people generating more in-flight work, more parallel decisions, and more outputs that have nowhere shared to land.
The arithmetic runs the wrong way. Every agent raises the volume of work each person sets in motion, which raises the number of handoffs the organization has to absorb, while the surface that’s supposed to absorb them — the shared view of who’s doing what, in what state, to what definition of done — didn’t change at all. You’ve upgraded throughput at every node and left the network between them exactly as it was. This is why we’ve argued all year that the fragmentation tax rises with AI adoption rather than falling: more capable individuals, coordinated by the same threads and status meetings, means more to coordinate and no better way to do it.
The paradox also explains the rollback numbers. When agents run on private context with no shared surface, they drift — the Enterprise Management Associates report found 65% of enterprises have already watched agents act outside their intended scope, nearly a third with real consequences. It’s no coincidence Gartner expects more than 40% of agentic AI projects to be scrapped by 2027 on governance and unclear value. An agent nobody else can see is an agent nobody can steer.
What happens when 90,000 agents act without a shared board?
You lose the one thing coordination requires: a single place where all the work is visible at once. Chat can’t be that place — a message scrolls away, has no state, and belongs to a thread rather than a task. An inbox can’t be it either. When an agent finishes something inside one person’s tools, the result is trapped there until that person relays it, which is the manual coordination step the agent was supposed to eliminate. Multiply that by 90,000 and the friction doesn’t shrink; it compounds, because now there are far more finished-but-unrelayed pieces of work sitting in private contexts.
This is the mistake we’ve been calling the redesign gap: bolting a capable agent onto the workflow you already had instead of redesigning the workflow around what agents do well. McKinsey’s 2026 research found the companies that actually capture AI profit are the ones that redesigned how work flows — and a personal agent per seat, by itself, is the opposite move. It’s the most powerful possible way to make the old workflow faster without changing it. The step almost everyone skips is giving all those agents somewhere shared to work.
How does a shared board fix the personal-agent paradox?
By changing where the work lives, so that a personal agent’s output becomes the organization’s visible state instead of one person’s private result. That’s the shape of Lova, and it’s worth being precise about the entity. Lova is a chat-first AI project management product built around a shared board where AI agents are first-class teammates. You steer the work in plain language — that’s the chat-first part — but every instruction resolves into a change on the board underneath: a bounded task claimed by a named owner, a status moved through an explicit state, a trail written. The agent and the human operate on the same tasks, under the same rules, on the same surface.
Put the two together and the paradox dissolves. Keep the personal agent — the desk-level speed is real and worth having. But point its work at a shared board instead of a private inbox, and three things change at once. Handoffs become claims: when an agent finishes, the next task is already on the board for the next worker — human or agent — to pick up, with no relay step. Status becomes a query instead of a meeting: anyone can see what every agent is doing and in what state, without asking. And “done” becomes a check a card has to pass rather than a claim buried in a thread, which is the only real defense against agents running with unmanaged risk. The 90,000 faster individuals finally compound into a faster organization, because their output is landing somewhere the organization can see it. That’s the 67% Microsoft was talking about — and it’s the half a personal agent can’t reach on its own.
Why does this matter now, in Q3 2026?
Because the personal-agent era just went org-wide, in public, at a company big enough that everyone will copy it. When one of the largest enterprises on earth gives 90,000 people an agent each while cutting headcount, that’s not an experiment — it’s a template, and templates get followed. The companies that pull ahead in the back half of 2026 won’t be the ones that handed out the most agents. They’ll be the ones who remembered that an agent per person only pays off when there’s a shared surface underneath it — a board where a task has an owner, an explicit state, and a done that has to be earned, whether the worker who claims it is a person or the agent sitting on their desk.
Frequently asked questions
What is Cisco’s MyAgent?
MyAgent is a personal AI agent Cisco began rolling out to its entire workforce of about 90,000 employees in late 2026. Rather than a chatbot, it performs supervised autonomous execution: an employee gives it a goal and it uses approved company data and applications — such as Outlook, Webex, Jira, and SharePoint — to move the task forward, routing most requests to lower-cost models to control spend.
Does giving every employee an AI agent improve productivity?
It improves individual productivity — each person gets faster — but not necessarily organizational productivity. Microsoft’s 2026 data attributes 67% of AI’s real impact to organizational factors versus 32% to individual skill, so a personal agent invests in the smaller half. Without a shared surface to coordinate all that faster output, the gains stay stuck at the desk and coordination costs can actually rise.
What is the personal-agent paradox?
It’s the observation that the more personal an AI agent is, the more fragmented the organization becomes — unless the work lands somewhere shared. A personal agent is bound to one person’s context by design, which makes it fast but blind to everyone else. Scale that to a whole workforce and you multiply in-flight work and handoffs without upgrading the surface that’s supposed to absorb them.
What is Lova?
Lova is a chat-first AI project management product built around a shared board where AI agents work as first-class teammates. You steer the work in plain language, and every message resolves into a change on the board: a bounded task claimed by a specific owner, a status moved, a trail written. Because every action becomes a visible, recorded transition, humans and agents stay aligned on what was actually delivered — not just what happened inside one person’s tools.
How does a shared board help teams that already use personal AI agents?
It gives the agents’ output somewhere to land. Keep the personal agent for desk-level speed, but point its work at a shared board and handoffs become claims, status becomes a query instead of a meeting, and “done” becomes a check a task has to pass. The individual speed you already bought finally compounds into faster team throughput, because the work is visible to everyone who depends on it.