People Are Vessels of Context and Agents of Action. So Is AirJelly.

“Man is but a reed, the most feeble thing in nature; but he is a thinking reed.” — Blaise Pascal

I have always loved this metaphor.

Human beings are fragile. We get tired. We forget. Our attention is painfully limited. Every day, we experience far more than we can hold on to.

And yet, we can look back. We can connect what happened yesterday with what matters today. We can understand the meaning of an experience, form a judgment, and decide what to do next.

Perhaps this is part of what makes us human.

We are not made by a single moment. A person is shaped by the books they have read, the people they have known, the promises they have made, and the decisions they have lived with. We are also shaped by unfinished plans, unanswered messages, and memories that return years later with new meaning.

We carry our past with us, and we use it to understand the present.

But a person is not merely a collection of context.

Hannah Arendt drew a distinction between what someone is and who someone is. Our roles, abilities, and résumés may describe what we are. Who we are, however, is revealed through what we say and what we do.

The past shapes us. Action allows us to keep becoming ourselves.

People are vessels of context and agents of action.

This is also how I think about AirJelly.

Why context

Today’s AI is remarkably intelligent, but most of the time it does not really know the person on the other side of the conversation.

Every new chat begins with an empty box. We explain who we are, what we are working on, what happened before, and what we need now. We copy in a meeting transcript, summarize a project, and gather fragments from documents, messages, browser tabs, calendars, and other AI conversations.

There is something strange about this.

Human collaboration becomes valuable over time because people develop shared context. Someone who has worked alongside you for years does not need a full briefing every time you say, “Let’s continue.” They know which project you mean. They remember why it matters, what has already been tried, and where things became difficult.

I increasingly believe that the next meaningful shift in AI will come not only from more capable models, but from continuity.

To understand someone does not mean recording everything about them. It means understanding the relationships between things: where a task came from, why a decision was made, how a project has evolved, who is waiting for a response, and which intentions were quietly buried beneath the work of another day.

That is my most persistent hope for AirJelly: that it can stay with a person across applications and across time, connecting the context that would otherwise become scattered or disappear.

Remembering is not enough

I have also come to believe that memory alone is not the destination.

If AirJelly simply saves everything, it may become a comprehensive archive—full of information about you, but unable to change what you are facing right now.

Context becomes valuable when it can be used.

Perhaps AirJelly brings back a discussion from months ago just before an important meeting. Perhaps it notices that a promise is about to be forgotten. Perhaps it connects several fragments of an idea you have been circling for weeks and helps you finally move it forward.

This is why I do not want AirJelly to be only a recorder.

I want it to understand context and turn that understanding into action. It should not always need to wait for a perfectly written instruction. It should be able to recognize what is unfolding, what remains unfinished, and what kind of help would be genuinely useful now.

To me, this is the essential step from context to agent: not knowing more for its own sake, but allowing what is already known to become useful.

Proactive, but never intrusive

“Proactive” is an appealing word. It is also, I think, the hardest part of what AirJelly is trying to become.

An AI that does nothing until it receives an explicit command will always remain a tool. But an AI that constantly guesses, interrupts, or makes decisions on your behalf does not feel like freedom either.

I want AirJelly to find the narrow path between those two extremes.

It should know enough about you to understand when it is worth appearing—and be restrained enough to know when it should remain quiet. It can prepare, remind, connect, and act, but it should not take away your judgment.

That is also why I care deeply about who owns the context.

A person’s work, relationships, habits, and memories are private and valuable. The context AirJelly understands should remain on the user’s own device, under their control: something they can inspect, manage, and delete.

I do not want AirJelly to become another system that tries to define you, or a machine that attempts to replace you.

I imagine it as a layer: something that sits quietly between you and the tools you use, helping you remember, helping you connect, and—when the moment is right—helping you act.

What I hope AirJelly can become

I hope that one day, we will no longer need to introduce ourselves to AI over and over again.

It will understand where we have come from, why the thing in front of us matters, and where we are trying to go. It will not reduce us to a profile or a collection of behavioral data. Through the context accumulated over time, it will gradually learn the unique rhythms, preferences, and intentions of each person.

I also hope it can carry some of the work that continuously drains our attention: searching for information we have already seen, reconstructing background, tracking loose ends, and remembering every next step.

Not so that people can do more and more forever, but so that we can reserve our limited attention for the parts that matter most: judgment, creativity, relationships, and life itself.

I do not know exactly what AirJelly will ultimately become. That uncertainty is part of what makes building it so compelling to me.

But I know that I do not want to create another AI that only answers questions.

I hope AirJelly can become an intelligence that shares a past with you and helps you move toward what comes next.

It carries context. It takes action.

Just as we do.

People are vessels of context and agents of action.

So is AirJelly.