Every project starts with a conversation. Someone says "we need to build X" and a discussion follows — scope, priorities, who does what. The conversation is where decisions happen. The board is where they get recorded, if anyone remembers to update it.
We've been asking: what if the conversation was the project management tool? Not a chat feature bolted onto a board, but a conversation that produces a board as a side effect.
The current model is backwards
Today's PM tools ask you to think in their structure first. Create columns. Create cards. Assign people. Set priorities. Then do the work. Then come back and move the cards. Then write an update so your manager knows what happened.
Every step after "do the work" is overhead. Necessary overhead, the tools argue — visibility requires documentation. But that's only true if the documentation has to be manual.
What if you just talked?
Imagine this: you open your project and say "We need to launch the new auth system. There's a research phase, an implementation phase, and a QA phase. Sarah handles research, Mike handles implementation."
The board builds itself. Three columns appear. Tasks are created and assigned. You didn't configure anything — you described what you need, and the tool structured it. If you change your mind ("actually, let's combine research and implementation into one phase"), you say so, and the board reshapes.
This isn't a hypothetical. It's how natural language interfaces work when they have enough context about the domain. The conversation is the input. The board is the output.
Status without status reports
The second half of this idea is about reporting. In a conversation-first model, status doesn't need to be written — it can be observed and synthesized.
When a team member moves a card to Done, that's a signal. When someone marks a task as blocked and specifies they're waiting on an API key from another team, that's a signal. When a task hasn't moved in three days, that's a signal too.
AI can watch these signals and narrate them back to the project lead in the conversation: "Auth implementation is blocked — Mike is waiting on API keys from the platform team. Research is on track, Sarah updated two tasks this morning."
No one wrote a status report. No one was interrupted. The lead knows what's happening because the tool is paying attention.
The interface that survives AI
Here's the part we find most interesting: a conversation-first interface gets more powerful as AI improves, not less relevant. Most PM tools are threatened by AI because their value is in the structure, and AI can generate structure instantly.
A conversation interface is the opposite. The better AI gets, the more it can do with your intent. Today it structures your board. Tomorrow it assigns tasks to AI agents who actually do the work. The interface stays the same — you're still just talking about what needs to happen. What changes is how much gets done automatically.
What we don't know yet
We're not sure this model works for every team. Large organizations with complex compliance requirements might need more rigid structure than a conversation can provide. Teams with dozens of projects might need dashboards that aggregate across conversations.
But for the startup team that just wants to track what's happening without drowning in tool overhead — we think conversation-first might be exactly right. The tool should feel like talking to a sharp colleague, not filling out forms.
We're building this now. Early results are promising: people use the tool more when interacting with it feels like talking instead of clicking. Whether that translates to better project outcomes is the question we're most excited to answer.