Microsoft has hired the full team behind Cove, a Sequoia-backed startup building an AI-first collaboration canvas, as the company intensifies efforts to reinvent how people brainstorm, plan, and co-create with Copilot. Cove is shutting down its product and deleting user data, with refunds issued and an export process offered, while its talent heads to Microsoft AI to work on what the founders describe as a larger vision for human-AI collaboration.
Inside Cove’s Bet on Canvas-Native AI Collaboration
Cove launched in late 2023 with a deceptively simple idea: chat alone isn’t how people think. Instead of linear prompts, it offered an infinite board where AI could place and rearrange cards, tables, lists, and timelines—creating a living plan rather than a transcript. Users could attach context from the web, PDFs, and images, and then nudge the AI to branch, refine, or consolidate as a project evolved.
The team was led by Stephen Chau, Andy Szybalski, and Mike Chu, veterans of Google Maps who helped ship Street View-era features known for blending information and spatial context. That pedigree showed up in Cove’s interface, which prized navigation, shared state, and fast iteration over long-form chatbot replies.
Backed by a $6 million seed from Sequoia Capital in 2024, with participation from investors including Elad Gil, Homebrew, Adverb, Scott Belsky, and Lenny Rachitsky, Cove targeted the crowded visual collaboration market. It competed with tools like Miro, TLDraw, and Kosmik, but leaned harder into agentic AI that could read materials, synthesize options, and restructure work on the fly.
Why Microsoft Wants This Kind of Talent Now
Microsoft has already threaded Copilot into its collaboration portfolio, including Whiteboard, Loop, Teams, and OneNote. What Cove built—AI that doesn’t just answer but actively constructs artifacts in a shared space—aligns with where Microsoft has been steering Whiteboard and Loop components. It points to meetings where boards are seeded with options before anyone joins, and to planning sessions where AI gathers materials, generates alternatives, and maintains a coherent project state as decisions change.
This is the frontier many enterprises care about: moving from assistants that summarize to systems that help teams produce. IDC forecasts that global spending on generative AI will surpass $140 billion by 2027, with collaboration and productivity among the fastest-adopting categories. The practical bottleneck now is workflow fit. Cove’s approach—multimodal inputs, spatial memory, and editable AI output—attacks that problem directly.
Microsoft has a track record of high-impact talent buys to accelerate its AI roadmap. In 2024, it hired key leadership and staff from Inflection AI to bolster model development and productization. Cove brings a different but complementary muscle: product designers and engineers who understand how to make AI feel co-creative, not merely consultative, inside a canvas where many people think visually.
What Cove’s Shutdown Means for Current Users
Cove informed customers that the service will shut down on April 1, with all user data slated for deletion. The company has refunded March subscriptions and is providing a data export process so teams can retrieve boards and assets. For organizations mid-project, a pragmatic path is to export artifacts and migrate to alternatives such as Miro or TLDraw for whiteboarding, while using knowledge management tools to store AI-generated materials for searchability.
The company has said the “ideas behind” Cove will live on at Microsoft. While there is no formal product announcement, the most natural landing zones are Whiteboard and Loop, where Copilot already assists with brainstorming and structure. Expect faster iteration on features like AI-driven board scaffolding, contextual search across attachments, and richer object types that can be manipulated by agents and people in real time.
The Bigger Story in AI Collaboration Tools
Generative AI started in chat windows because that was the easiest on-ramp. But complex work rarely fits a single thread. Product roadmaps branch, research needs citations and files, and planning requires versions, not verdicts. That is why “canvas-native AI” is gathering pace—from design tools that propose layouts to boards that draft itineraries, org charts, or sprint plans as manipulable objects. Cove crystallized that shift; Microsoft now has its builders.
The strategic prize is not only better brainstorms. It’s durable, machine-readable structure: cards, dependencies, sources, and decisions that AI can reuse across apps. If Microsoft can make those objects portable—Whiteboard to Loop to Planner to Teams—Copilot becomes more than a helpful narrator. It becomes the connective tissue that remembers context, proposes next steps, and maintains continuity across meetings and milestones.
What to Watch Next as Microsoft Integrates Cove’s Team
Keep an eye on rapid updates to Whiteboard and Loop that make AI output more editable and board-aware, and on Copilot features that ingest PDFs, web content, and images directly into shared canvases. Also watch for more seamless handoffs between visual ideation and structured execution, such as exporting AI-curated boards into tasks and timelines without losing context.
For Microsoft, this is a compact bet with outsized leverage: hire a team built around a clear thesis—AI works best when it thinks on a canvas—and wire that thinking into the tools hundreds of millions already use at work. For Cove’s founders, it is a chance to scale their ideas with deeper resources and a larger surface area. For users, the message is straightforward: the product is winding down, but the concepts that made it compelling are likely to reappear where they can have far greater reach.
