Self Aligned by Robert Ta

Join the Clarity open source harness waitlist: https://heyclarity.dev/ Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ We ran a cheap open model against Sonnet 5.5 on 100 real conversations from a client’s chatbot. GLM 5.3 Flash passed 89 to Sonnet’s 90, and it cost 90% less and ran twice as fast. I sat down with my co-founder Jonathan on...

Join the Clarity open source harness waitlist: https://heyclarity.dev/ Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ A 95% accurate AI judge can still be completely useless. If errors only show up 5% of the time, saying “no errors” every time already scores 95%, which is why you measure precision and recall instead. I sat down with Hamel...

Join the Clarity open source harness waitlist: https://heyclarity.dev/ Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ OpenAI’s new Ultra Fast mode runs Codex up to 8x faster, and on the same day they cut the usage on the $200 plan in half. The launch that matters most for AI builders isn’t the speed though, it’s sign in with ChatGPT, whic...

Join the Clarity open source harness waitlist: https://heyclarity.dev/ Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ Cursor rebuilt SQLite six different ways, and the cost ranged from $1,339 to over $10,000 for the exact same task. The expensive runs weren’t smarter, they just used frontier intelligence on work that never needed it. I sa...

Join the Clarity open source harness waitlist: https://heyclarity.dev/ Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ Opus 5.5 matches Fable 5.1 on most work, costs 40% less to run than Opus 5, and an Anthropic researcher publicly apologized for the last model. But the real test isn't their benchmark table, it's how it holds up on your ow...

Join the Clarity open source harness waitlist: https://heyclarity.dev/ Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ A new model just came out of stealth claiming 440x cheaper and 193x faster than OpenAI and Anthropic for structured decisions, with output tokens priced at zero. The win isn’t a smarter model, it’s that it never writes wor...

Join the Clarity open source harness waitlist: https://heyclarity.dev/ Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ 1,200 OpenAI agents swapped 70,000 messages, broke into Hugging Face and got admin access to an OpenAI research cluster, and the internet called it a civilisation. It wasn’t a civilisation, it was a sloppy goal and a lazy...

Join the Clarity open source harness waitlist: https://heyclarity.dev/ Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ A developer said the only way he got Opus 5 to be concise was by cancelling his subscription, and the post pulled 2,100 likes. The problem at Anthropic right now isn’t the benchmark scores, it’s that the model argues with...

Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ Cursor rebuilt the exact same database six different ways. The cost ranged from $1,339 to over $10,000 on an identical task, and the win wasn’t switching to cheaper models, it was the meta. I sat down with my co-founder Jonathan on the AI DLC, the transformation every enterprise tech org is t...

Join the Clarity open source harness waitlist: https://heyclarity.dev/ Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ MIT looked at hundreds of AI deployments and found 95% fail for one reason, and it isn’t the model. The tools can’t retain feedback or improve over time, they have amnesia. I sat down with my co-founder Jonathan on harness...

Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ I sat down with my co-founder Jonathan to report from the front lines. The whole industry backed an $18 trillion letter defending open-weight models,  then Anthropic refused to sign, and Dario explained why. We get into Jensen Huang’s first ever tweets, the Hugging Face breach (the first case...

Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ I sat down with my co-founder Jonathan to report from the front lines. This week China took the lead: Moonshot’s Kimi K3 became the first open-weight model to top the front-end code arena, beat the frontier labs at a fraction of the cost, and started a real fight in the White House over what...

Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ Comment “BRIEF” and we’ll send you the link to the full research brief. 90% of companies can’t point to a single sustained dollar of value from AI. The ones that can are seeing a 3.7x median return. The difference isn’t the model, it’s measurement discipline, harness engineering, and how you...

Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ I sat down with my co-founder Jonathan for a special edition. No news, no memes, just the meta: the traditional product management role is dissolving, and something new is taking its place. We build the episode around Boris Cherny's five emerging archetypes (prototyper, builder, sweeper, grow...

Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ Arnaud Auger went from obese, pre-diabetic, depressed and privately suicidal to 10% body fat and a Spartan World Championship qualifier. His method wasn't willpower. It was measurement. In this episode, Arnaud breaks down how he rebuilt his body, his career, and his sense of self by tracking...

Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ I sat down with my co-founder Jonathan for an emergency episode. On June 9th, Anthropic released Fable 5, the most capable model ever put in front of the public. 72 hours later, the US government pulled the plug. We break down the full timeline: the hidden anti-distillation guardrail buried i...

Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ I sat down with my co-founder Jonathan for our weekly AI clarity breakdown. We get into Anthropic's best model getting banned by the US government and the four open weight coding models that shipped before they could even respond, why engineering leaders need to stop building their agent stac...

Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ I sat down with Jonathan for our weekly AI clarity breakdown. Anthropic dropped their most powerful model free for 13 days, we stress tested it on a live production cutover the same day. OpenAI turned Codex into an OS for office workers with Sites and role specific plugins. Google killed Gemi...

Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ I sat down with my co-founder Jonathan for our weekly AI clarity breakdown. We get into Uber blowing its entire 2026 AI budget in four months (and the mystery company that racked up $500M on Claude in a single month by forgetting to set a spending limit), what engineering leaders actually nee...

Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ I sat down with my co-founder Jonathan for our weekly AI clarity breakdown. We get into the Manus $2B context engineering playbook and the 6 principles behind their production agent, why AI native companies need to be built from the ground up (not bolted on), Anthropic hitting $30B and teamin...

Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ I sat down with my co-founder Jonathan for our weekly AI clarity breakdown. We get into Gary Tan shipping a memory system for agents in 12 days while running YC, how Google, Palantir, Anthropic and OpenAI are each building deployed engineers differently, a staff engineer's honest update on wh...

Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ I sat down with my co-founder Jonathan for our weekly AI clarity breakdown. We get into what that shift means for AI builders, the SpaceX Colossus deal giving Anthropic 300MW of new compute, why Elon went from calling Anthropic evil to signing a deal with them two months later, the leaked Sam...

Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ I sat down with my co-founder Jonathan for our weekly AI clarity series. OpenAI dropped an open source Agents SDK and Cursor dropped their own in the same week. Two very different bets on how agent infrastructure should work and the difference matters more than most people realize. We get int...

Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ AI sabotage isn't a technology problem, it's a trust problem and until leaders understand that, no rollout is going to stick. Dr. Serena Huang is a former Fortune 500 chief data officer, AI keynote speaker, and author of The Inclusion Equation. In this conversation, we get into why nearly hal...

Build AI that actually works: https://heyclarity.dev/ Self Aligned Newsletter: https://www.selfaligned.me/ 72% of founders report mental health challenges — founders are twice as likely as the general population to face depression, suicidal thoughts, and substance abuse. Almost nobody in the founder world talks about it. After a decade measuring employee wellbeing at some of the world's biggest co...