The Last Thing for Humanity to Do · Harshal Gajjar / blog<br>Harshal Gajjar<br>Harshal Gajjar is an AI Forward-Deployed Engineer at C3 AI, based in the San Francisco Bay Area. Harshal leads Agentic AI harness development for the Forward-Deployed Engineering organisation at C3 AI, and since January 2026 has been building a stealth-mode startup in the Agentic AI space. Harshal cofounded Shram.io in 2024, where he led the pivot from a Jira-competitor product to an AI assistant that reached #2 Product of the Day on Product Hunt.<br>Harshal holds an M.S. in Computer Science (Machine Learning specialisation) from Georgia Tech and a B.Tech in Computer Science from IIT Dharwad, where he was part of the institute's foundational class. He spent three summers at Wolfram Research in Boston — first as a summer researcher in 2018, then as an instructor for high-school students in 2019 and 2020 — and was a Wolfram Student Ambassador throughout his undergrad.<br>Outside of work, Harshal is a long-distance cyclist and a vertical and horizontal caver, active with the San Francisco Bay Chapter (SFBC) grotto. In 2019 he was part of the Hubballi Bicycle Club Guinness World Record for the longest single line of bicycles.<br>Contact Harshal at mail@harshalgajjar.com.
piece·Jul 7, 2026
The Last Thing for Humanity to Do<br>AI made doing free, then deciding. The scarcity keeps climbing — and it's not clear the last rung stays ours.<br>There's an old puzzle about why water, which you'd die without, costs almost nothing, while a diamond, which does nothing for you, costs a fortune. The answer is that price never really tracks how much a thing matters. It tracks how short of it you are. Water is everywhere, so nobody pays for the next glass; diamonds are rare, so people pay through the nose for the next stone. Important and expensive quietly come apart, and once you notice that, you start seeing it everywhere.<br>You especially start seeing it in the way people talk about AI.<br>For as long as software has existed, the expensive part was the doing — actually sitting down and building the thing. Engineering hours ran out before anything else did, so that's what the whole market learned to price: the salaries, the headcount, the job itself. Building was the water we were perpetually short of, and being short of it was the entire reason it cost what it did.<br>Deciding what to build sat on the other side of that, more like the diamond. Not because it didn't matter — a wrong call could burn a year of expensive engineering — but because you were never actually short of it. There were always meetings to be had, planning docs to write, whole quarters to think it over, and most of that judgment sat idle anyway, waiting behind the engineering queue for its turn. Deciding was high-stakes and abundant at the same time, which is exactly what water is, and so, like water, nobody thought to put a price on it.<br>The natural objection is that if doing was so expensive, then deciding must have been at least as expensive, because getting the decision wrong cost you that whole year. But that quietly folds together the two things water and diamonds spent all that effort pulling apart. A wrong decision was high-stakes. It was never scarce. And what a thing costs follows how scarce it is, not how much rides on getting it right.<br>Then AI drains the water. Once the doing is nearly free — once one person ships what used to take a team, and a working prototype is a lunch away — the thing you're suddenly short of is good decisions, made quickly. Deciding didn't become more important than it always was; it just became scarce for the first time. The old skill was execution: could the thing even be built. The new one is aim: which problem is actually real, because the machine will happily build the wrong thing at superhuman speed and gift-wrap it for you.<br>You can watch this play out in the numbers, too. When AI got good, output per engineer went up, headcount came down, and valuations wobbled. The easy story is that AI took the jobs, but that same stretch is also when cheap money dried up, and the layoffs fit that just as well. AI didn't so much pull the trigger as remove the excuse. For a decade "we need more engineers" went unquestioned, and most of that demand was for raw doing, which is precisely the part that just went free.<br>So AI didn't make judgment valuable. It drained the water and left us staring at what we'd been paying for the whole time. It will build anything you can specify; it still can't tell you what's worth specifying. The doing turned out to be water, and deciding turned out to be the diamond it had been hiding all along.<br>But there's an obvious next question, and it's the one that actually keeps me up. What happens when AI gets smart enough to aim well too — to decide as well as it already builds? If it drains the diamond, where does the scarcity go?<br>It doesn't stop. It just climbs another rung. Because the whole thing is a regress, and each rung was only ever in service of...