Jake

@chrisbeetweets I think the shift will lean towards the ideation layer less towards execution. I think the larger question will be if that increases tools needed two fold or reduces them.

Something is shifting on the product side. Specs are starting to matter in a way they didn't before. The teams leaning into that are moving differently. Curious what others are seeing right now. Where do you think this is in a month?

I read recently that only 31% of product leaders feel confident they're building the right thing. Honestly? That number feels high. We have more data, more AI tools, more feedback channels than ever before. And most of us are still making gut calls and hoping the next sprint backs us up. The job hasn't changed that much. We just have fancier ways of being uncertain now.

My current take as having worked in AI for some time now is I'm not worried about it taking developer jobs. Honestly, I've never been more excited about what's ahead for people who decide to lean more technical. The barrier to building real things just collapsed. Someone who would have spent years learning to code before shipping anything meaningful can now get there in months. That's not a threat, that's an opening. High level languages made code 10-100x faster to write. Developers were supposed to disappear. Instead we got the largest software boom in history. AI is just the next abstraction layer. Binary didn't kill developers. Assembly didn't. Python didn't. The hiring freeze isn't AI. It's COVID overhiring and nobody wants to tell their shareholders they got greedy and made a mistake. The scope of software about to be built will dwarf everything before it. The jobs are coming. If you've been sitting on the fence about going deeper into tech, this is the moment. Not in spite of AI. Because of it. "This time is different" is the most recycled wrong opinion in tech.

The spec surviving contact with execution is the whole game. As a PMM, I've written briefs that felt airtight and watched agents build something technically correct that missed the point entirely. The problem isn't the spec. It's that nothing forces the original intent back into the room when it actually matters. You can't politely ask for good output. You have to make bad output impossible.

So, it's been a bit of an interesting journey the last couple of weeks. On Feb 5, I started building an app that would let me tell agents to build things (code) while I play a round of golf. Which is funny because, I mean, it's winter in Boston, so I'm not exactly out on the links on the weekends. That said, it gave me a good (if somewhat douchey) framework to figure out what I think the future of engineering could look like. Now I'm 150+ sprints (and 1,500+ commits in) and have 298 documented "oh, that's why that broke"/"gotcha" moments, and ended up what looks remarkably close to something that might actually be useful, lol. Everyone knows by know that agents can write good code. They just also have amnesia, and they're prone to ignore suggestions in the name of efficiency. If you make a rule, it must be enforced. Claude's hooks help with this. But otherwise, they'll build the wrong thing faster than you could imagine. So I started treating them less like people and more like highly skilled Leonard Shelbys (for the Memento fans). Short sprints (two hours, not two weeks). Gates that block bad merges instead of prompts that politely ask for good ones. Everything written down, because nothing survives in their memory. Of course, I'm calling it Agentile. It's Agile, but for a team that has to reconstruct their reality every time they wake up in the morning. It's just context, stupid! Write-up here: lnkd.in/exVemukp

@artman @linear This is sick. Reviews are where context goes to die. Anything that keeps the loop tight (diff, CI, convo) is a win. We’ve been focused on the stuff before the PR so the review is boring in the best way.

Happy to share that code reviews are coming to @linear, available in private beta on every plan. Modern code reviews with structural diffing that vastly reduce the number of changed lines in many cases. Review, comment, check previews, get notifications on failed CI check. And finally merge the PR, all from within Linear. Comments are synced two-ways with Github.

@ika_xbt People aren’t ready to break 6 bera is programmed for minimum 25$

@RiceRiddler Proud to be included in both ecosystems both are so sick

@0xFent @henlo Henlo

@deepname99 Beras have always been prepped for our part

@whoiskevin The last rebase is a job at the McDonalds franchise @deepname99 buys with his bera airdrop

The last rebase is one of the best kept secrets, ever. Through all of these years, not once, have I ever heard any alpha about it.

@justjay Hope the surgery goes well man, ooga booga <3

Wut r the mfering odds I have surgery on mainnet…

@0xDrewBeefy Mibera

It’s not enough that I should succeed - my friends should as well. - Mibera

@henlo [ ]

@deepname99 henlo

@deepname99 @henlo Been the CEO

@deepname99 @0xhoneyjar Been booga before

@henlo Ooga booga

@Tma_420 Always

Spread the Bera love

@Bera_Punks bm

Henlo.

@henlo Cultivating berachain