How do you find your feet at a young AI agent company? Based on my experience at AirJelly, you might need to meet four requirements.
First, you need to be young enough.
I mean your outlook and your mind, rather than your age. Stay curious about the world. Be willing to investigate things you have never seen before. Feel compelled to click on a new product. How many tabs you open afterward, and whether you still remember what you originally meant to do, can be discussed later.
Second, you need to be physically strong enough.
As someone who exercises regularly, I have adapted well to the environment here. Looking at all the different brands of protein powder and creatine on the desks, it is hard not to wonder whether a résumé needs at least a year of gym experience to make it through screening. People talk about agents and training alike. Mental and physical work each have their own supplements.
Third, it helps if you understand a state of mind that feels a little like ADHD.
You are working on one thing while three more appear in your head. New ideas keep coming, your to-do list keeps growing, and your attention occasionally takes a business trip of its own. You do not have to have ADHD to find this familiar. Before long, you may have a simple wish: could something help me keep hold of all these things and move them forward a little?
At that point, you understand why AirJelly exists.
Finally, you need to be smart enough, and kind enough.
There is a high level of trust in this young team, and everyone is responsible for a lot of things. When we talk, people are thinking about how to do the work well. You rarely need to run an extra program in your head to analyze what someone “really meant” by a remark. With fewer unwritten office rules to navigate, there is more mental space for the problems that actually need solving. It is a comfortable way to work, but it asks a lot of you, too: understand the problem quickly, take action, and take responsibility for what you have agreed to do, so colleagues can confidently leave the next step with you.
Of course, none of these are official hiring criteria. Proof of your gym history is not currently required. Once I actually joined, I found that the process of becoming part of the team could be described through four questions.
1. Should you understand the product first, or finish the task first?
My first assignment as a General Affairs Specialist at AirJelly was to research the proactive agent products currently available.
Step one: work out what a proactive agent is.
Step two: find relevant products, compare their features, and put together a report.
My method: open AirJelly and have it work through all of those steps with me.
Yes, I used AirJelly to research AirJelly’s own field. It even told me, “That was a solid step.” I had only just joined, and the product itself had already given me its approval. After looking around, I found few products among the supposed competitors I examined that directly matched AirJelly’s core positioning. I also began to understand the problem it aims to solve: proactively identifying and moving forward the next step from the context of desktop work, reducing the burden of organizing background information and issuing instructions.
The background to our work often already exists in the pages, documents, and tasks we are looking at. If every request for AI help starts with gathering that background again, explaining how things got here, and describing what we want it to do, even preparing the question takes time. For a team that needs to move quickly, that preparation deserves attention too. The research assignment itself became my way of getting to know the product. From defining the concept to compiling the report, I was completing the task while learning about the product helping me complete it.
So, understand the product first or finish the task first? In practice, I did both together. Start using it with a real problem, then turn that understanding into something you can deliver.
2. How do you find your place when things change quickly?
Most assignments a General Affairs Specialist receives are just one sentence long. The task is brief; the background needs to be pieced together across departments.
My information comes from a wide range of sources. When a task arrives, I usually need to find the relevant colleagues and understand what each of them knows. Fortunately, people are very cooperative and willing to provide plenty of detail. I rarely find myself thinking, “Nobody told me.” More often it is, “Everyone has told me. Now it is my turn to put the pieces together.”
Once I have collected the information, I need to clarify four things quickly:
- What needs to happen now;
- Where the work is getting stuck;
- What resources and support we are likely to need;
- How much time it is likely to take.
I then report back to the person who assigned the task with those assessments, giving us a basis for planning the next steps and coordinating resources.
Here, execution shows up in many small, specific actions: getting started promptly, seeking people out when context is missing, flagging obstacles early, and making resource needs clear. “I’ll follow up” should be followed by action, progress, and a result. If a problem cannot be resolved yet, the people involved still need to know where things have stopped.
This is how I gradually found my place: take a one-sentence request, turn it into clear actions, and keep following through until I can give a clear account of the outcome.
Being reliable means colleagues do not have to keep wondering what happened to the thing they handed over to you.
3. What changes when your colleagues all use agents?
When colleagues make extensive use of their own agents, gathering information across departments becomes faster, too.
People have accumulated plenty of information in their daily work. When I need their help, they can ask their AI to organize the relevant parts and send me something comprehensive and detailed. I ask a question, and they can quickly come back with the information already pulled together. As the person responsible for combining it all, I welcome this. These materials help me understand the situation faster and work out the actions, obstacles, resources, and timing. They also reduce the back-and-forth of follow-up questions and missing context.
When everyone uses their own agent, the benefits spread through the people they work with, shortening the time it takes to move the whole task forward. The time saved can go toward judgment and action: which information still needs checking, which problem should come first, and who needs to work with whom on the next step. Agents speed up the work of organizing information. I remain responsible for the assessments I pass on and for what happens afterward.
Of course, general affairs work brings another ongoing challenge: the tasks may not all be big, but every one needs to be remembered. One needs following up today, another needs checking in a few days, and some require no action now but must come back to mind at the right time.
I place cautious trust in my memory. So whenever something has a deadline or a specific time attached, I tell AirJelly right away: remind me to do this at that time. After all, there is still considerable room for improvement between “Got it, I’ll remember” and actually remembering when the time comes. For me, setting reminders promptly is part of taking responsibility for a task. The more things I take on, the more I need a way to keep following through, so each commitment comes back into view at the right moment.
4. When did you first feel you belonged?
I cannot point to a particular moment and say, “That was the day I became part of the team.”
I simply began to notice that people assumed I knew about more and more things. Colleagues came to me to ask where things stood, and trusted the information and assessments I gave them. As a General Affairs Specialist who often has to go around asking other people for context, becoming someone others turn to for context made belonging feel tangible. That trust is direct, and it carries weight. When people plan their next steps around the information you provide, you need to distinguish carefully between what is confirmed, what is still being checked, the expected timing, and any changes that need to be shared promptly. When people are willing to trust you, you want each response to deserve that trust.
Outside work, there is another readily observable measure of belonging: how often people bring you food.
Everyone may work out, but that does not stop all kinds of high-calorie food from appearing in the office every day, and eventually making its way to me. Protein powder and creatine on the desk coexist quite naturally with a snack just handed over by a colleague. Training plans and plans to keep colleagues well fed both run steadily here. This added another piece to my understanding of the team: people take work seriously, training seriously, and feeding their colleagues seriously too.
Looking back, becoming part of AirJelly happened through these small, concrete things: using the product for my first research assignment, turning a one-sentence task into a plan that could move forward, using an agent to keep track of scattered information and to-dos, and gradually becoming someone colleagues were willing to ask and trust. Moving from understanding the product to helping build what comes next takes curiosity, quick action, consistent follow-through, and responsibility for results. Everyone on the team carries a lot. Taking on part of that work, and giving others confidence that it is in good hands, is a very real way to contribute.
As for the gym requirement I mentioned at the beginning, nobody has formally tested me yet.
But given how much food comes my way each day, I had better keep training.