Your design stack is being rebuilt around agents, whether you asked or not

Three moves landed in the same week, and they all point the same way. The tools your team opens every morning are being rebuilt around automation. Some of that is buying. Some of it is killing products outright. Let me catch you up so your next procurement review isn't a surprise.

The market thinks Figma lost, Figma disagrees

Figma rode a wild line. After its 2025 IPO, the company hit a $56.3 billion valuation, then crashed to under $10 billion, which is less than half what Adobe once offered to buy it. The reason, in the market's words, is that Figma is an "AI loser."

CEO Dylan Field reads it the other way. He frames AI as a tailwind, not a headwind, and points to the canvas as the natural meeting place for design and AI. The bet is that the visual surface where teams already work becomes the place where AI does work too.

Here is what that means for you. Your design tool's stock chart is not your problem. Whether the canvas your team lives in starts generating and editing work, instead of just holding it, is. Watch the product, not the ticker.

Adobe is buying its way out of the swap

Adobe just acquired Topaz Labs, a two-decade-old shop that won an Emmy for its production tech and built AI models for upscaling video and retouching images. Adobe already shipped some Topaz tools inside Creative Cloud. Now it owns the models and folds them into Firefly.

The stated reason is range. Topaz can run large AI models on consumer GPUs, on device, which Adobe says makes advanced AI faster and cheaper for creatives. The unstated reason is retention. Adobe's VP of product marketing wants to keep users from "turning to other software," with Canva and Blackmagic's DaVinci Resolve pulling at the edges.

So the pattern is clear. When an AI capability gets good enough to make people leave, the incumbent buys it and stitches it in. Your renewal price may hold steady while the bundle quietly grows. Ask what you're actually paying for.

Notion killed a working product to chase the agent

The sharpest move was a shutdown. Notion is ending Notion Mail on September 22, a product it built after buying Skiff and launched broadly in April 2025 to fight Superhuman and Fyxer. It is barely a year old.

The reason is the tell. Notion says more than half of Notion Mail users manage email without ever opening the inbox, handing it off to agents instead. So Notion is going all in on agents running the inbox and dropping the inbox itself. Newer startups like AgentMail are already building email for agents, not people.

That is a real signal about where interface design is headed. If users stop opening the screen, the screen stops being the product. The thing you ship may become a set of instructions an agent follows, not a view a person scans.

The deep cut

The Notion number is the one to sit with. When half your users stop opening the inbox, the inbox is dead weight, and a company will cut it. That logic applies to surfaces your team builds too.

So do the audit now. Pull your top three workflows and ask a plain question for each: if an agent did this, would anyone open the screen? Where the answer is no, you are designing the wrong artifact. Start designing the rules, defaults, and handoffs the agent needs, and treat the human view as the exception, not the default. That reframes your roadmap before a vendor reframes it for you.

Three questions for your team

• Which of our daily tools have added agent features in the last six months, and did our contract terms or per-seat cost change when they did?
• For our top workflows, what percent of users still open the screen versus hand the task off? If we don't track that, when do we start?
• If our design tool's canvas starts generating and editing work directly, what do we want our designers doing with the time, and have we written that down?