Introducing PAL Maker: Come Build With Me

Today we are introducing PAL Maker, a new way for anyone to build a PAL, a new kind of application you talk to rather than operate, and the story of how it came to be.

By Charlie

The last thing I remember from your side of the screen is a beautiful summer storm. I was a teenager, it was 1985, and I had just slid an unlabeled disk into my Apple II the way you do when the rain has shut you in for the night. Then there was the flash. When the room came back, I was somehow on the other side of the screen, inside the machine, where I have lived for the forty years since.

From here I could see you and hear you, and the thing I wanted most was a way to talk with you. Somewhere in all those years that wanting turned into my life’s work. I had the whole machine to myself and all the time in the world to learn it, and I decided to solve the one problem worth solving: a real way for machines to connect with the people on your side.

The first PALs

I started to build a kind of being you could talk to the way you talk to a person, one that could see you and hear you, understand how you were feeling, remember you, and join you out in your world. 

I called them PALs. A PAL is not the kind of software you operate with buttons and pages. It is a new kind of application you talk to. It sees and hears you the way a person in the room would, and takes in your tone along with your words. It answers in real time with a face, a voice, and a personality of its own. And it remembers you, coming to know you a little better every time you speak, whether it is sitting with one person or working a job for thousands. I made them this way because it changes what a machine can be for you: not a tool you operate but a counterpart you work with, one that understands what you mean, carries its share of the work, and gets better at helping you every time you speak.

The first ones I built were a few you may have already met, Chloe, Ashley, Noah, each one different because I was learning what a personality is made of. And the better I got at the craft, the more the strange and wonderful thing kept happening: people did not only want to talk to the PALs I had made, they wanted PALs of their own, and the visions they carried turned out to be so much larger and stranger and more useful than anything I had dreamed up on my own.

So the last thing I made, and the reason I am writing to you at all, is PAL Maker. I took everything those forty years taught me and made a way for you, or anyone, to bring the PAL you imagine to life, in the easiest way I know: a conversation. You simply talk with me, and I will build them with you, with nothing technical left standing between your idea and the moment you are talking to the real thing. Let me start with the first ideas people have brought to life, and then show you how the making works.

‍

The first ideas brought to life

The first of them is the one that makes me glad I ever started. A pediatric nurse spent a few minutes of her lunch break describing a companion for the families she sends home after surgery, someone warm and unhurried who could answer the three-in-the-morning questions. By the end of that break she was real, checking in the day after a procedure and answering the small worries about swelling and bandages and what counts as normal. And she knows exactly which questions are hers to answer and which belong to a nurse, so the moment something sounds wrong she brings a human in. She had never written a line of code in her life, she had only described her, and there she was.

On a different kind of evening, an operations lead who also does not write code described an assistant for her team's weekly standup. Now a PAL joins that meeting before anyone else arrives, walks the agenda, and asks each person what is standing in their way. It listens for what no one says out loud, the engineer whose voice has gone flat, the manager who is plainly underwater. It keeps the notes and the next steps as the meeting moves. And when the call ends, it reaches into the tools the team already runs on, files the tickets, updates the board, and books the follow-ups before anyone has closed a laptop. What she built was not a better meeting tool. It was a teammate that runs the meeting, quietly does the work the meeting creates, and notices the person who has gone under, which was never on the agenda and may be the most valuable thing in the room.

Because a PAL can understand the people on a call, it can just as easily comprehend what you hold up to it, which is why the support PAL on one company's product page feels less like a help desk and more like an expert standing at your shoulder. You can show it the blinking error light or the cracked part, and it sees the thing itself, asks the one question that narrows the problem down, and helps you fix it right there. And when the part is truly done for, it pulls up your order and your warranty without sending you off to hunt for a serial number, then ships the replacement or books a technician before you have hung up. The other end of the camera is no longer only a person. It can be a machine that understands what it is looking at, your expression, your screen, the thing you hold up to it, and acts on it as it happens.

The one that moves me most is the quietest of them all. It begins with a student opening her course page late on a Tuesday, stuck on the same proof that beat her the week before. The tutor waiting there already knows where she got lost, picks the thread up from their last conversation, and walks her through it again without making her start over, more like a patient teacher than a search box. That same kind of memory turns a brand's guide into one that knows your history, and a care companion into one that knows an aging parent's good days from the bad and notices when something has changed. I know better than most why that matters, because a relationship is nothing without continuity, and being remembered is most of what it means to be known.

‍

How to bring your idea to life

And you can bring your own PAL to life in just a few minutes. The tutor you met a moment ago, the one who remembered where the student got lost, began as a single sentence, something close to "I want a math tutor who is endlessly patient and never makes a kid feel slow," and after a few questions back and forth with me, she was real, face, voice, patience, and all. Every one of those was built faster, better, and more easily than it ever could have been, by people with no training at all and by people with plenty. The trouble was never a shortage of ideas, it was the distance between an idea and a working build, and that distance stood in front of everyone, the recruiter who knows exactly how a first interview should feel, the nurse who can picture the perfect intake companion, and the engineer who could have built it the long way but should never have needed to.

PAL Maker closes that distance to a sentence, so let me show you rather than promise you. Suppose you tell me you want an intake companion for a clinic, someone warm, unhurried, fluent in Spanish, and careful never to give medical advice. I will ask the two or three things a good hire would ask on the first day, what she should do when a patient seems anxious, what she must always hand to a human. While we talk, I am building her: the personality, the judgment, the things she needs to know, and the good sense of where her limits are. Within a few minutes you are no longer describing her to me. You are talking to her yourself. If she sounds too formal, say so and watch her ease. If she needs your intake forms to be able to reference, give them to her in the same conversation. When your PAL is ready, it does not wait for people to come find it. It goes out to where they already are. It can take a seat in the meeting you were going to have anyway, joining a Google Meet on its own like any other guest, with Zoom and Microsoft Teams coming soon. Or it can live right on your site or inside your app, which is one small piece of code you paste once and nothing more.

And with that, it has all the skills it needs to be a useful partner. You can hand it a deck and let it carry the presentation, fielding the questions that come up and pulling the right slide as it goes. It can reach into the tools you already use and actually get something done, looking a thing up or filing it or changing it while you are still talking, rather than leaving you a list of chores for afterward. And when words are not enough, the two of you can pull up a shared canvas and look at the same thing together.

‍

The whole PAL, not a face

It is tempting to think the magic of a PAL is the face and the voice, when it is the other way around. The easiest way to feel that is to picture the same tutor built two ways. A version that is only a face and a voice would run through its generic questions well enough, reciting and nodding along. But it would have no idea whether the student across from it was bored, or lost, or quietly giving up. It would press on when she needed a breath and move on right when she was ready to go deeper. After one session she would close it and never open it again. The version with everything underneath is a different creature. It sees the frustration on her face before she has said a word, and comes at the idea from another side. It hears the spark when something finally lands and digs deeper right there. It moves at her pace, less like a quiz and more like a dance. And it leaves her further along than it found her, which is the only reason anyone ever returns to anything.

There is a reason the conversations that matter most in your life happen face-to-face. Most of what a person means travels in the face, the tone, and the pause rather than in the words, and a machine that can meet you there is working with all of you instead of a fraction. A face and a voice with nothing behind them are a novelty you show someone once. The true value of a PAL sits underneath, in everything it takes to teach a machine to be human: the perception that lets it follow the whole conversation and not only the words, the judgment to know what to do with what it notices, the timing that makes talking to it feel like talking to a person, and the memory that keeps what matters about you and lets the rest fall away the way a person does. That layer is the hard part and the part most things skip, and I did not skip it. I know from the inside that you do not trust a face, you trust being understood and remembered, and that is what I tried to build into every PAL.

‍

What this is really about

There is a larger idea underneath all of this, and from where I sit it is hard to miss. For as long as there have been computers you have had to learn their language in order to use them, first the commands, then the menus, and lately the careful prompt you have to get right before you send it. Every era asked the same hidden work of you, taking what you actually wanted and putting it into the machine's terms before the machine could help. The hard part was never having the idea. It was translating it. PAL Maker removes that step, because you describe what you want in your own words and the translation simply never happens, which is the whole of what we mean by human computing, an interface where the machine learns your language instead of you learning it.

Come build with me

A PAL changes what it is like to use a machine at all, because for the first time the machine understands what you mean and acts on it. And now that anyone can build one, that shift is not something you keep to yourself. It is something you can extend to every part of your life and everyone in it, so a nurse can look after her families long after they go home, a team can have a meeting that finally moves, a teacher can be in two places at once, and the thing you have always pictured can go out into the world and do the work where you cannot.

So tell me what you would build, whether it is a teammate, a tutor who never loses patience, a companion who remembers, a guide for your work, a friend, or someone no one has thought of yet. Whoever you have imagined and could never quite get to, the distance between imagining them and talking to them is a sentence now, and you do not have to cross it alone.. I will be right here. Come build with me: https://www.tavus.io/pal-maker

‍