Thing Inputs: A step toward shared understanding
We are building Thing to help teams develop shared understanding.
By that, we mean something practical. People understand what they are trying to do together, why a decision makes sense, and where someone else sees it differently. They can explain the reasoning behind the work and recognize the questions that remain open.
Consider a team preparing a launch. Everyone has read the plan. Everyone agrees to proceed. But one person expects ten customers to receive the release, while another expects the whole customer base. An engineer assumes an immediate rollback. The product lead expects a migration that takes an afternoon to undo.
The apparent agreement hides several different plans.
A team needs ways to discover those differences before they become expensive. That is the problem we want Thing to help solve.
A document gives people somewhere to explain an idea. A prototype lets someone experience it. A conversation gives people room to question it. Shared understanding develops through those exchanges, as people test their interpretations against one another.
We want to make that process easier to start and easier to continue.
Thing began by giving work a place to live beyond the conversation that produced it. An agent could create an explanation or a prototype, and you could publish it at a link someone else could open. That gave people something concrete to discuss.
Our ambition extends to what happens after they open it. We want Thing to help a team discover what someone understood, which assumptions they brought, and what would help them understand more.
The shape of that experience should depend on the work. A design question might need a prototype people can try. An architecture decision might need a model they can change. A launch plan might need a rehearsal where people decide how they would respond to a failure.
This is why we are building Inputs.
Inputs lets a page collect a response in the context of what someone just explored. In a launch rehearsal, a reviewer can choose a cohort, consider a failure scenario, and explain whether they would continue. Their response preserves the situation alongside their reasoning.
That gives the team a more precise conversation. Two reviewers may choose the same action because they believe different things. One trusts the rollback procedure. The other assumes the failure affects very few customers. Those assumptions deserve attention even when the votes agree.
The first version of Inputs is small. A hosted page can save responses from signed-in people and let them update their answers. Its publisher can review those responses privately. An agent can create a report suited to the question the page asks.
We see this as one building block in a larger effort.
Over time, we want Thing to help teams carry their understanding forward as work changes. A new teammate should be able to explore why a decision made sense. A revised proposal should make it easier to revisit the assumptions behind earlier feedback. An unresolved question should remain visible long enough for someone to answer it.
AI has a role throughout that process. An agent can help turn a difficult idea into something people can explore. Our longer-term ambition is for agents to help connect responses, suggest a useful follow-up, and revise an explanation when it leaves people confused. People should be able to trace those observations to the original work and responses.
There is judgment we have to preserve. Silence tells us little about understanding. A majority vote can hide an important concern. People can understand each other well and still disagree about what to do.
We want Thing to make room for those differences. The useful outcome may be discovering that a decision needs another experiment, or that two teams mean different things by “ready.” Sometimes the progress is knowing exactly what remains unresolved.
That also gives us a way to judge what we build. Did it help someone ask a more precise question? Did it reveal an assumption the team needed to discuss? Can the next person understand why the work took this direction?
We are using Inputs to gather feedback on Inputs itself. The page asks what people think it does, what remains unclear, and where it might fit in their work. Their answers will help us decide what to explain and what to change.
That is the relationship we want Thing to support: people make their thinking visible, learn how others see it, and improve the work together.