From 10x people to 10x teams
The first wave of AI was single-player. The next wave is multiplayer.
The last 18 months happened fast.
Teams that hadn’t touched AI in early 2024 now use it every day to generate ideas, write PRDs, analyze research, and build working prototypes. I see it in my own work. I went from using AI a few times a day to running three or four threads in parallel, all day long.
Output is up. Velocity is up. Everything looks faster.
And yet, when I sit with product teams — startups, scale-ups, and a few household names — I keep seeing the same picture. Clogged GitHub repos. Stacks of unread AI-generated PRDs. Prototypes in various states of completion. Features abandoned by ambitious individuals who never got the buy-in to get them across the finish line.
We gained something powerful. We also lost something important.
AI has turned the most ambitious people on every team into 10x versions of themselves. But it has also made it harder than ever for them to play together.
Everyone’s doing context engineering in their own silo — generating work that’s overlapping, repetitive, and untethered from the rest of the team.
Last week, I attended Miro’s Canvas 26 event in San Francisco. This is exactly the problem Miro is trying to solve — and that makes Canvas 26 worth a close look on two levels. If you use Miro, it shows how the product is evolving to amplify how your team collaborates with AI. If you don’t, it’s just as worth a look, because it points to a larger shift in the industry: from siloed, single-player context engineering to collaborative, multiplayer context engineering.
That shift matters.
The future of AI-powered product work won’t be defined by who can generate the most artifacts. It will be defined by who can turn shared context into shared progress.
A quick note on how this piece came together: I was at Canvas 26 in two roles — as a speaker, closing the day with Brian Balfour in a fireside chat on how AI is changing teams, and as a Miro creator partner for the event. Miro sponsored this post; the analysis, the opinions, and what I chose to focus on are entirely my own. Learn more about Canvas 26 ↗ · @Miro · #MiroPartner #Canvas26 #sponsored
The canvas is the context
In April 2025, Brian Balfour and I launched Reforge’s AI Strategy Program. Jeff Chow, Miro’s CPTO, has been a guest in every cohort. That has given me a unique window into how quickly Miro is moving — both in what the platform can do and in how Miro’s own team works to ship it.
What stands out is not just the velocity. It’s the throughline.
Every leap is rooted in the same principle: The canvas is the team’s shared context.
It is the place where messy human collaboration becomes visible. Where ideas, decisions, prototypes, and tradeoffs can live together. Where the work stops being trapped in individual heads, docs, chats, and tools.
That is the real shift: People using AI alone → Teams using AI together.
Miro’s bet: make AI multiplayer
At Canvas 26, Miro unveiled a set of features designed to help teams work in an AI-native way on a surface they already understand: the collaborative canvas. The announcements build on Miro’s existing AI foundation, but push the product in a more agentic direction — toward a world where AI is not a tool the team reaches for from time to time, but a companion that works alongside the team.
The most important shift: Sidekicks
Sidekicks have evolved from AI assistants into agentic thought partners. They can pull context from across the team’s stack, ask clarifying questions, break a problem into steps, and generate output in the formats teams actually use: diagrams, docs, sticky notes, kanban boards.
A chat assistant produces a discrete artifact — you ask, it answers, the work ends. A Sidekick is different. It lives on the board, takes on multi-step problems, and keeps improving the team’s work as new context comes in. It’s the same shift you feel going from Claude in a chat window to agentic tools like Claude Code or Claude Cowork — where the model can take a goal, run with it, and operate inside your actual environment. Same underlying intelligence, dramatically more powerful harness.
My favorite feature: Flows
Every product team eventually hits the same wall: a hundred sticky notes on a board and no clear path forward. Flows are built for that moment. They let teams turn messy inputs — workshop notes, customer quotes, rough ideas, half-formed concepts — into structured outputs ready for the next stage of the product development lifecycle.
Not by sending agents off into the background. By keeping the human in the loop.
The best AI workflows don’t replace judgment. They concentrate it.
The coolest technical feature: Code to Prototype
You can now have a live, working frontend prototype sitting directly on your Miro board. Until now, prototypes lived somewhere else: in Figma, in code, in a demo link, in a handoff doc, in someone’s memory. The team would click through a static mock and then argue, in the abstract, about how the real thing should behave. You can’t debug a hover state when the hover state isn’t there to debug.
Putting the prototype on the board closes that gap. The team can interact with the actual thing, argue about the actual interaction, refine the actual flow.
The artifact stops being downstream of collaboration. It becomes collaboration.
The most foundational feature: Miro’s MCP server
MCP connects the canvas into a two-way flow of information. Context from a board can be pulled into an agentic coding session, and that session can turn around and manipulate the board itself. The canvas becomes readable by AI. And eventually, writable by AI.
It’s the glue that connects collaboration in Miro into the tool stack necessary to build, ship, and operate exceptional products.
The most human feature: Custom Widgets
Teamwork is messy, colorful, and fun. Custom widgets keep the canvas a space built for humans — not just one optimized for AI agents.
When building gets cheap, choosing gets harder
At the end of the day, Brian Balfour and I sat down for a fireside chat on how the role of product builders is changing. For a long time, the role was fundamentally about prioritization. We could only build a small part of our roadmap, so we had to be incredibly good at it. The central question was:
What should we build?
That question still matters. But AI changes the weight of it. When teams can generate concepts, prototypes, copy, flows, and even working code faster than ever, the harder question becomes:
What should we ship?
What earns a place in the product? What improves the customer experience? What adds power without adding complexity? What do we say no to, even though we can now make it?
The role shifts from prioritization to curation. And curation is a team sport.
You cannot curate alone. You need product, design, engineering, sales, and customer success looking at the same plan, with the same context, at the same time.
This is exactly the gap Miro’s latest features are built to close. The dozens of sticky notes from a workshop don’t sit there forever — they get shaped into something the team can actually define, build, and evaluate together.
Miro has a clear-eyed vision for how AI integrates with your team. Taken together, the announcements at Canvas 26 aren’t just features. They’re designed to give teams the blueprint to build a new AI-native way of working.
The 10x team era
The first wave of AI made individuals faster. The next wave will make teams faster.
The real bottleneck in product work isn’t the speed of producing artifacts. It’s the speed of shared understanding — getting the right context into the room, aligning on what matters, making the hard calls, and carrying those decisions forward.
Teams that build shared context will compound: better inputs lead to better outputs, faster decisions, and sharper context for the next cycle. Teams that stay siloed will keep producing artifacts nobody else can build on. They’ll mistake motion for progress.
The next decade of product won’t be defined by 10x individuals. We already have those. It will be defined by 10x teams — teams that turn scattered expertise into shared context, and shared context into compounding execution.
That’s what Canvas 26 is really about. Not a new set of features, but a new way of working.
See what changes when the whole team is finally in the same room — and the AI is in the room with you.
Disclosure: I’m a Miro creator partner for Canvas 26. Opinions are my own. Learn more about Canvas 26 ↗ · @Miro · #MiroPartner #Canvas26 #sponsored
Interesting. I've been trying to use Miro via its MCP from my cursor agent which had access to our product knowledge base. I've found it woefully inadequate, unable to generate a basic swimlane diagram with very specific guidance. To me, visual artifacts from AI are still a big weakness. Nano banana keeps creating the same lame AI designed content, Miro doesn't seem to work properly. I've reverted to using Claude cowork which itself has a lot of issues but it's the best I've found so far. I really want Miro to be better though. We love using it and I still make manual visuals on it.