Why the Chat Interface Is Beating Dashboards for AI

A few years ago, the consensus was clear: AI would dynamically generate custom interfaces for every interaction. Buttons, sliders, dashboards — assembled on the fly, perfectly tailored to the task. It sounded obvious. It was also wrong.

Jay Hack, Head of AI at ClickUp — the $4B work management platform where non-technical teams run their entire workflow through AI agents — has watched this prediction collapse in real time. After building Codegen, one of the first coding agent companies, and integrating it into ClickUp’s platform serving marketing firms, construction companies, and legal teams, he has a counterintuitive read on why chat keeps winning.

The Prediction Everyone Got Wrong

The smart money said AI would assemble a custom interface for every back-and-forth. Dynamic buttons. Interactive charts. Tailored dashboards. And to be fair, some of that has happened — Claude generates inline HTML graphs, and it works well. But as the primary interaction model? It never took hold.

“The reason that chat has been so successful is because humans vastly over-attribute the amount of work that they will be performing in the future,” Jay explains. “They won’t actually be viewing a dashboard to understand where the data is going and make a decision on that. They’ll in fact be delegating the viewing of that dashboard to an agent as well.”

The implication is stark. If you’re building a product around showing users data, you’re building for a behavior that’s disappearing.

Delegation Changes Everything

The logic follows a chain that most product teams haven’t internalized yet. If an agent can view a dashboard, interpret the data, and come back with a recommendation — why would the user view it themselves? The agent just says “I think you should increase pricing” and the human decides yes or no.

This maps directly to how delegation already works between humans. An analyst doesn’t hand their boss a raw spreadsheet and say “interpret this yourself.” They synthesize, recommend, and present. AI does the same thing, except faster and without the politics.

“I actually do think chat has an incredible amount of staying power,” Jay says. “Pretty much any cognitive task that you can draw a circle around and say, this is something that a human performs — we will probably get to the point that an AI will perform it. And that implies that the chat interface is going to have a lot of staying power.”

The Presentation Layer Doesn’t Disappear — It Gets Cheaper

Jay does carve out one exception. Presentations — the artifact humans use to communicate reasoning up a hierarchy — still matter. A mid-level manager still needs to explain their reasoning to leadership. But the cost of producing those artifacts has cratered.

“Even though it might not be the first thing that you end up doing, I think there will still be a place where essentially your higher up says, ‘can you please explain to me why we’re doing this thing?’ And then you can very quickly spin up an agent who will document all of that reasoning.”

The shift isn’t from dashboards to chat. It’s from humans consuming data to humans reviewing agent recommendations. The interface that supports that workflow is conversational, not visual.

What This Means for Builders

If you’re building AI-powered tools today, the takeaway is uncomfortable: complex UIs may actually be a liability. Every custom interface you build is a bet that humans will continue performing the cognitive work that interface was designed to support. Jay’s experience across both Codegen and ClickUp suggests that bet is getting worse by the quarter.

The products gaining traction are the ones that let users delegate in natural language and get back a recommendation, not a visualization to interpret. That’s not laziness — it’s the same pattern humans have always used to scale their decision-making. We’re just extending it to software.

FAQ

Why are companies moving from dashboards to chat-based AI interfaces?

Most dashboard usage involves a human interpreting data and making a decision. AI agents can now perform both steps — reading the data and recommending an action. Chat becomes the natural interface because users delegate the cognitive work rather than performing it themselves. The shift mirrors how human delegation already works in organizations.

What types of teams benefit most from chat-based AI tools?

Non-technical teams — marketing, legal, accounting, construction — benefit most because they never built dashboard-reading into their core workflow. ClickUp’s primary customers are these non-technical knowledge workers who spend all day in the platform and prefer delegating tasks through natural language over learning specialized interfaces.

How does ClickUp’s AI agent handle complex multi-step tasks?

ClickUp’s agents operate across all workspace data — tasks, chats, documents, whiteboards, and meeting notes. A user can ask an agent to pull cost data from Datadog, check Salesforce pipeline, compile findings, and notify the finance team, all through a single chat thread without switching between applications.

Can chat interfaces handle tasks that require visual outputs?

Yes. HTML, JavaScript, and CSS have become effective output mediums for LLMs. Claude already generates interactive charts inline. The difference is that the AI generates visuals when needed rather than requiring users to navigate to a separate dashboard. Jay notes that “as LLMs become superintelligent, maybe they won’t even need to show you a chart.”

What is the risk of building a product around custom AI-generated interfaces?

The risk is building for a behavior that’s declining. If users increasingly delegate both data interpretation and decision-making to agents, complex custom interfaces become unnecessary overhead. Products that require users to actively interpret visual data are betting against the delegation trend.

How does the chat interface model scale for enterprise collaboration?

In multiplayer environments like ClickUp, chat threads carry shared context across team members and agents. An agent that compiles a financial analysis can post findings in the right channel, tag relevant stakeholders, and maintain conversation history — replicating how human teams already coordinate through messaging.

Why did custom AI interfaces fail to gain traction?

The prediction assumed users would continue performing cognitive work but wanted better tools for it. Instead, users preferred to offload the work entirely. Dynamic button and slider interfaces solve an optimization problem for a task the user would rather not do at all.

Is the chat interface trend specific to AI or broader?

It reflects how humans have always communicated to get work done — through conversation, not dashboards. Text messaging, email, and Slack already dominate workplace coordination. AI chat extends the same pattern to human-machine collaboration, making it feel natural rather than requiring users to learn new interaction models.

What should product managers prioritize when designing AI-first tools?

Focus on the recommendation layer, not the visualization layer. Build for the interaction where the agent delivers a synthesized answer and the user makes a binary decision. Jay’s experience suggests that product and engineering roles are converging around the chat interface, collapsing the toolset gap between Figma and IDE users.