Forward Deployed Hospitality

As AI gets better, building gets cheaper.

An idea that once needed a designer, an engineer, a project manager, and several weeks of work can increasingly become a working prototype in an afternoon. That changes where the difficult part lives. When almost anyone can build, the advantage shifts upstream, toward noticing what a person actually needs, understanding why they need it, and describing it clearly enough to make something useful.

I think of that skill as forward deployed hospitality, the ability to get close to a person, sense what would improve their experience, turn those signals into something buildable, and stay around long enough to learn whether it worked.

For most of the software era, execution carried much of the cost. You could understand a problem well, have a strong idea, and still be unable to solve it because you lacked the time, money, or technical team. Plenty of good ideas died somewhere between intention and implementation.

AI shrinks that distance. It can draft an interface, write code, analyze feedback, document a system, and produce several variations before lunch. Implementation still matters, of course, but a bad premise can now travel much faster. Misread the human need, and you can build the wrong thing with impressive speed. Write a vague specification, and you are likely to receive a polished interpretation of your vagueness.

The scarce work, then, moves closer to deciding what deserves to be built, for whom, and why.

People rarely hand you a clean requirement. They say, "This dashboard is confusing," "The room feels off," "Onboarding takes too long," or, "I wish this felt more personal."

You can treat each sentence as a direct instruction, rearrange the dashboard, change the lighting, remove a few steps, or put the customer's first name on a welcome screen. You will have responded, although you may not have understood much.

The dashboard user might be afraid of making an irreversible decision. The room might feel off because nobody knows where to stand when they arrive. Onboarding might feel long because the user cannot see any progress, or any reason to finish. Personalization might have little to do with a name, and much more to do with remembering context.

The request tells you where to look. It rarely tells you exactly what to build.

Finding the need underneath it takes attention, curiosity, and judgment. You have to notice pauses, repeated complaints, workarounds, body language, emotional stakes, and the gap between what people say and what they consistently do. At some point, you also have to make a bet about what those signals mean.

Hospitality often gets flattened into a few familiar instructions: smile, remember a name, offer a drink, be nice. Warmth matters, but great hospitality is unusually precise.

A good host notices that someone is cold, and offers a blanket before the guest has to ask. They can tell whether a surprise will create delight, or anxiety. They understand that a solo diner needs a table, but may also need some evidence that they are welcome there. They tell a guest what shoes to wear because the beautiful route includes soft grass.

Care shows up in details, which makes hospitality a useful form of requirements gathering. A thoughtful host keeps converting human signals into decisions about timing, environment, information, and action.

Product work follows a similar path. You notice a signal, form a view about the human need beneath it, decide what feeling or outcome would be better, write a specification, ship the change, and watch the response. AI is becoming very capable in the middle of that sequence, especially when the outcome and context are clear. The beginning and the end still require someone to perceive the signal, make sense of it, and judge the result.

Forward deployed engineers work near the customer, and inside the environment where the software has to perform. A tidy requirements document is rarely waiting for them. They enter a messy workflow, learn the constraints, ship a narrow solution, watch it operate, and carry what they learn back into the product.

Hospitality benefits from the same proximity. A conference room can tell you only so much about a human experience. Watch someone use the product, sit in the waiting room, walk the arrival path, listen to the support call, and try to find the bathroom without asking. Notice when someone hesitates, apologizes, abandons a task, or invents a workaround.

Then put care into that exact environment, through a better default, a clearer instruction, a graceful recovery, a remembered preference, or one well-timed human intervention. The work is forward deployed because the truth is usually in the field, sitting inside ordinary behavior.

Observation, listening, context, and synthesis all gain value here. Can you see friction that people have normalized, hear the need behind a request, learn the language and incentives around the problem, and combine scattered signals into a plausible explanation of what is happening?

Taste and specification matter, too. A technically correct result can still feel awkward, overbuilt, or strangely indifferent. Someone has to choose among the possible solutions, then explain the chosen one with enough clarity that another person, or an AI system, can build it.

Judgment matters most when the signals conflict. Which need deserves attention? How much intervention is proportionate? Where should automation stop? When should a person step in? The answers depend on the customer, the context, the consequences, and what happens after launch.

Follow-through closes the loop. Stay close, observe the consequences, and revise the original theory when reality disagrees. These skills are sometimes called soft, although they decide whether all the hard work creates value.

Combining AI, personal context, and hospitality carries an obvious risk. The same tools that help us remember can help us manipulate. Anticipating needs can become an excuse to collect everything, and personalization can become a pleasant-looking machine for extracting more attention, or more money.

My rule is simple: collect only what you can use to improve the person's experience, and use it in a way they would be comfortable understanding. Thoughtful hospitality increases agency, removes friction, clarifies choices, preserves dignity, and helps someone get what they came for. Once intimacy becomes leverage, the care is gone.

AI makes it easier to practice either version, which leaves the choice with us.

Before building, get close to the experience, collect the requests, and pay equal attention to hesitation, repetition, workarounds, emotion, and context. Name the need beneath the symptom, then write down the change, its boundaries, and what success would look like.

While building, use AI where it genuinely compresses the work, and deploy narrowly enough that an imperfect theory will not damage the entire experience. After shipping, watch what improves, what breaks, and what people do next. Keep the useful parts, revise the weak ones, and turn the lesson into a repeatable capability without sanding away the judgment that made it useful.

This extends the idea behind Service by the Senses: observe more inputs, connect them to human context, and turn care into operations.

We'll probably get much more competent output, more software, content, interfaces, automations, and immediate answers, delivered at very little marginal cost. As the volume rises, specificity becomes more valuable.

Who is this for, what are they trying to do, what are they afraid of, and what friction have they stopped noticing because it is so familiar? What would make the experience feel functional, calm, and considered?

Knowing how to use AI will matter. Knowing what to ask it to make, whom that work serves, and how to tell whether it helped will matter more.

The valuable skill is noticing someone accurately enough to make the right thing for them. That is forward deployed hospitality.