
Spoiler: ever seen those neat cable management videos on the internet? That's what engineers do. In their heads. Agents can't do it for you, because your head is still yours :)

Founder of @codespeak_dev | Was in charge of creating Kotlin

Spoiler: ever seen those neat cable management videos on the internet? That's what engineers do. In their heads. Agents can't do it for you, because your head is still yours :)

If coding is solved, what's left? My talk at Agent Engineering HQ is not about software factories or the latest tricks of Claude Code. It's about engineering. Because that is the future. Take a look :) youtube.com/watch?v=pDliDj6s…


I thought @claudeai UIs (TUI, VSCode, Web, Desktop, Mobile) were embarrassingly buggy, and then I tried Codex 😱 Remote just doesn't work most of the time. Opening an existing session in desktop app is a dice roll: maybe it is there, maybe not. Features are randomly missing from one UI or another (Claude has this too)... Use @codespeak_dev, people, at least you'll be able to have the same requirements across all your apps. Brave new world 🤦♂️

Introducing Unreal Agent: An open-source harness with state-of-the-art cost efficiency 39% cheaper than Codex+Astra on Terminal-Bench 4.0 while maintaining performance


🤩 Vote for my talk at Serving Code Awards: sirviendocodigoawards.com/?n… Thanks in advance!


meet @mostik_ai! what happens when you put 12 PhDs in one room for four months? first place on the ARC-AGI leaderboard, which I can't say much about while the competition is still running. and this, which I can. everyone's arguing about whether open models will catch up to frontier models. we think it's the wrong question. here's the one we pose: why does a frontier model have to generate your answer at all, when the only thing you need from it is the reasoning? we do this by enabling models to communicate in latent space. through our protocol, hidden states pass straight from a frontier model into a small one running on your infrastructure -- no text between them, and neither model is fine-tuned. two models from different families, sharing reasoning, both left untouched. how do we know it works? we tested it on a setup where a 753B model reads the problem, and a 4B edge-class model writes the answer. with this approach, we get results 80% as accurate as the frontier model, but at 20x faster performance. we're committed to preventing frontier model lock-in and are already partnering with inference providers to accelerate open-weight adoption. we've done this between 15 of us, in four months, 12 PhDs and a Fields medalist, backed by @generalcatalyst WIRED has the first external account of the company and the work: wired.com/story/russian-star… full writeup, the setup, and all the numbers: mostik.ai/read-more

In the future, humans won't be writing much code by hand, but this does not mean that 𝗦𝗼𝗳𝘁𝘄𝗮𝗿𝗲 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴 is going away. We've never been paid for knowing the syntax of a programming language or a specific API. We've always made money on building 𝘀𝘆𝘀𝘁𝗲𝗺𝘀 𝘁𝗵𝗮𝘁 𝘄𝗼𝗿𝗸 𝗮𝗻𝗱 𝗸𝗲𝗲𝗽 𝘄𝗼𝗿𝗸𝗶𝗻𝗴 𝗳𝗼𝗿 𝘆𝗲𝗮𝗿𝘀. Maintaining such systems take a lot of engineering. But what 𝘪𝘴 engineering? And what parts of it will be left when coding machines get even better than they are now? Spoiler: it's not as bad as many people think... youtube.com/watch?v=r3GyLrg0…

Had a great chat with @thegeeknarrator about the future of coding, developer productivity, @codespeak_dev, and how the world is changing: youtube.com/watch?v=JgYMH7nB…

Hundreds of CodeSpeak users taught us a simple truth: 𝗻𝗼 𝗼𝗻𝗲 𝘄𝗮𝗻𝘁𝘀 𝘁𝗼 𝘄𝗿𝗶𝘁𝗲 𝘀𝗽𝗲𝗰𝘀 𝗯𝘆 𝗵𝗮𝗻𝗱. So we re-thought the whole thing. Now we extract your intent from your chats with agents and turn it into 𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲𝗱 𝘀𝗽𝗲𝗰𝘀 behind the scenes. More details and a waitlist are here: codespeak.dev/blog/spindle-2…

If you have 5 mins, please help us 𝗹𝗲𝗮𝗿𝗻 𝗺𝗼𝗿𝗲 𝗮𝗯𝗼𝘂𝘁 𝗿𝗲𝗮𝗹-𝗹𝗶𝗳𝗲 𝗮𝗴𝗲𝗻𝘁𝗶𝗰 𝗰𝗼𝗱𝗶𝗻𝗴 by filling in this survey: l.codespeak.dev/agentic-codi… Want to have a more in-depth 1:1 chat with the CodeSpeak team about your agentic coding practices? Please tick a box on the form above 🙏

A short bit on Intent Recovery at AI:AM last week. Also talking with my hands A LOT, apparently 🤌 youtu.be/MnaLvzhYekg?si=11md… - why current agentic coding is a bit backwards - how @codespeak_dev can help - what we can and cannot know about the future Thanks @labenz

⚠️ 𝗖𝗵𝗮𝗻𝗴𝗲 𝘁𝗵𝗶𝘀 𝗖𝗹𝗮𝘂𝗱𝗲 𝗖𝗼𝗱𝗲/𝗖𝗼𝗽𝗶𝗹𝗼𝘁 𝘀𝗲𝘁𝘁𝗶𝗻𝗴 𝗡𝗢𝗪 𝘁𝗼 𝘀𝘁𝗼𝗽 𝗱𝗶𝘀𝗰𝗮𝗿𝗱𝗶𝗻𝗴 𝘆𝗼𝘂𝗿 𝘄𝗼𝗿𝗸. Tell your Claude Code: Set `cleanupPeriodDays` to 3650 days (10 years) 𝗔𝗻𝗱 𝘆𝗼𝘂𝗿 𝘄𝗼𝗿𝗸 𝘄𝗶𝗹𝗹 𝗯𝗲 𝘀𝗮𝗳𝗲.

Experience report on using @codespeak_dev, and specifically `takeover`: towardsdatascience.com/i-let…