aphel

@michael_chomsky It’s an obvious product, because it’s fundamentally useful. Devs have wanted it since Stripe popularised good docs. I tweeted about it in 2021 at least (x.com/itsaphel/statu…) It’s not luck, it’s just creating real value. And there’s tons of it to go around

mintlify has got to be the luckiest company I have ever seen in my entire life >ships markdown first docs just as LLMs cause an explosion in developer products >gets into YC, and turns out most of the batch needs docs >wait, other YC companies also need docs >other doc products are built for C-suite and are unusable for scrappy fast-moving developers >other docs products couldn’t grow past a point so there’s not a lot of scrappy competition. the industry looks dead and non-venture scale. >Agents are a thing, and turns out having docs in your codebase is no longer a good-to-have DX thing, but a necessary 10x unlock >searching docs is hard, so Mintlify acquihires the entire Trieve team (that previously built the fastest end-to-end RAG product in Rust. I was thinking about joining Trieve at some point bc I was so impressed by it when I was working at a vector db) >now Mintlify has the best search of any docs product by far >search quickly drives revenue as companies are willing to pay good money to have effective chatbots in their docs >turns out docs have a crazy k-factor, because developers actually see and spend a long time on docs. Mintlify doesn’t even need to parade their logo—their customers are selling their product for them! >refocuses on UI/UX/Latency and now security >but eventually that early bet that let them succeed became a limiting factor. SOMETIMES I WANT A FUKIN EDITOR THAT SYNCS WITH GITHUB >Mintlify builds a world-class editor for docs, and also agents for docs >at this point fast growing AI startups have noticed mintlify and are migrating like crazy >big legacy enterprises carefully eyeing these AI cos start migrating too. mostly INBOUND?????? Yes there’s a lot of truly world-class execution here, but lots of truly lucky bets tbh at ended up going insanely well for them. Mintlify is 2-3 good decisions from being a decacorn. I wouldn’t bet against the Hidden Leaf of YC

My predictions for 2026 "Everything works now and that's insane" Coding is, for all intents and purposes, solved in the next generation of models. Everything related to coding that isnt “solved” is more of a harness/context problem than anything else. This means everything from backend, reliability and deployment, security, and frontend is able to be handed off to an agent. The exception is actual business logic and user experience, which still doesn’t feel right till two more generations of visual and spatial reasoning improve. Browser agents become highly effective and are used for the long tail of problems that cant be solved by purpose built software. As a result of this, bot protection becomes a really big thing people have to fight/worry about to get this tooling to work and companies like browserbase make a shit ton of money doing so.Everything from booking a flight, getting a dinner reservation, to scraping leads on linkedin and applying to jobs for you is doable with a browser agent. The browser won’t be the place people access this - it’ll just be in a normal chatbot style interface that’s async. There’s a breakthrough in “continuous learning” in the first quarter, and a new paradigm labs call “learning models” (or something similar) start getting released every quarter. Anthropic gets there first, then openai follows. Continuous learning is a combined system of evolving context and continued finetuning on a model. This doesn’t solve the context engineering problem, but does effectively eliminate the need for finetuning entirely. Agents get really good at managing their own context and context of sub-agents around them. Because of this, multi-agent swarms start to work really well. Context expansion, context compression, and retrieval become a solved problem by the end of the year the same way structured output generation is now a solved problem. AI designing interfaces with things like the figma MCP or just writing straight to code get really good. We enter uncanny valley of design in early-to-mid 2026. AI assistants move to single threaded and away from having a ton of different chats. Single threaded is the way people interact with other people, and was a difficult problem pre-2025. But, since evolving context and memory became mainstream, single threaded experiences like Poke dominate. This is in prep for voice mode to become one of the default ways to interact with AI. Voice mode gets a big step up around the middle of the year. Pre-2026, voice modes were just odd and still feel like talking to a robot. Greatly improved voice modes make voice a default way of interacting with the models. “Her” is basically reality by fall. AGI, if defined as Samantha from the movie “Her” (my personal definition) is a reality by October 2026. It’s a single threaded experience but you can access context in a very clever and guided way. OpenAI claims they have it first, though for various legal reasons might not actually say its AGI, and it’s released as a consumer product. There is much debate over whether this is actually AGI. The ending of the movie where the OS’ all band together to make ASI is not happening in 2026. OpenAI IPOs cause they need the capital. They release “Chat-1” or whatever they wanna call it during the roadshow. Nano-Banana-67 and similar new tooling basically automates investment analysts out of their jobs; it still takes a long time to diffuse because of tooling and organizational lag. Google becomes the leader in AI for biology, and acquires some biotech companies to deploy it. Basically no human trials are seen, though the company wants to run trials on people in developing countries to accelerate time to market and this causes some PR issues for approximately 48 hours on twitter. Apple leads the way in proposing a new standard to verify images and videos are not AI generated, and starts requiring social media apps on the app store to follow this standard. It uses camera and microphone metadata to prove something came directly from a camera. They get buy in from Google, Tiktok, and Meta by the end of the year by basically forcing them to do it. We see some seriously insane game demos from the world simulation companies, but none of it is really playable until early 2027. Sim gaming exists, and kinda works, but its not great yet. Similar vibe to early VR. Cerebres or Groq gets acquired by one of the labs. Microsoft acquires Cursor for some absurdly high number. Govt gets scared of China, so they give OpenAI and Anthropic 100 billion dollars or something crazy like that (long shot but wanted to put this down anyways). Apple buys Thinking Machines after Anthropic turns them down. E2B or Daytona (the agent sandbox companies) gets acquired by Cursor. Waymos are in every major city except New York (they’ll never work in new york). Waymo has a problem with manufacturing enough cars to meet demand. A few founders go to jail for lying about ARR numbers. Seed round valuations decline from current levels but the AI hype doesn’t die down too hard. (Some of these are more thought through than others. also published to my personal blog)

@rvbhadja @lgrammel @aayushkapoor_ smolagents by @huggingface can effectively do something similar. smolagents use a custom python interpreter to intercept function calls. I suspect Anthropic is doing something similar internally

@lgrammel @aayushkapoor_ great job! any idea how to archive this with other providers for example openai?

Super interesting feature, the Anthropic code execution provider tool can directly call client tools. Great implementation by @aayushkapoor_

Wonder if long-term this decision should be made at the application layer. For some use-cases, it's deterministically clear which tool results are likely not useful anymore. (and doing so in code is cheaper than using LLM intelligence)

Annoying that there's not a standardised template for ASAs in the UK, like there is YC's SAFE in the US. There's an assorted set of templates that differ in details that seem significant. Probably an opportunity for a good law firm to free publicity by releasing one.

.@natfriedman: "I just fundamentally don't believe the world is efficient"

An API we integrate with completely broke an endpoint. They then sent a notification that the endpoint will be deprecated in 2 years. No plans to fix it in the meantime. APIs man…

The @ilyasut episode 0:00:00 – Explaining model jaggedness 0:09:39 - Emotions and value functions 0:18:49 – What are we scaling? 0:25:13 – Why humans generalize better than models 0:35:45 – Straight-shotting superintelligence 0:46:47 – SSI’s model will learn from deployment 0:55:07 – Alignment 1:18:13 – “We are squarely an age of research company” 1:29:23 – Self-play and multi-agent 1:32:42 – Research taste Look up Dwarkesh Podcast on YouTube, Apple Podcasts, or Spotify. Enjoy!

@tatuya01 ngl the AI brainrot is quite entertaining

the best ai slop embraces that it is slop

@levelsio I don’t think the UK is too bad atm. London’s startup ecosystem is doing increasingly better

I just realized the only 2 countries left with actual substantial startup activity now are literally only the US and China The rest of the world can't really do startups, doesn't have the funding, can't grow them and it's more like performative hobby projects for their governments Which might tell us where the future wealth will be concentrated in the world

Sometimes I feel that there are two different Anthropics. One that does great research and products, and one that generates PR articles that have nothing behind them...

Spent the better part of this year working on hardened sandboxing, and I’d agree. Never found anything in open-source or literature that was quite fitting

Glad people see through the bait more quickly now

@theo I mean, that’s an abuse (and the regs have a way to deal with this), but the regulation itself is definitely good for consumers & businesses Tons of scummy stuff and broken product subscriptions out there. This reg lowers risk and therefore barrier to purchasing

If you ever wonder why some companies doing support EU customers, I hope this helps explain

@ChrisMurphyCT You're being played by people who want regulatory capture. They are scaring everyone with dubious studies so that open source models are regulated out of existence.

@blakeir Summary notifications are a decent solution now. Batches the cruft together

Always be on Do Not Disturb with emergency bypass on. Nothing is that urgent. 99% of notifications should be default off anyways.

here's my gripes about mcp as an implementer - end users expect miracles from mcp and load in a bunch of them - most mcp servers are garbage, crash a lot, and the llm never bothers calling them - this whole "try sse and if it fails try streaming" is exactly why im so anti standardization when things are early, already a ton of bloat - even if an mcp is good 99% of people won't know to configure it - it's now on me to create some complex ui that helps people manage all their mcp tools - companies felt pushed to offer mcp and have half assed implementations that users expect to work well it really feels like the kind of product dev i hate - come up with some idealistic vision of an interconnected seamless future, create a spec out of it, everyone gets hyped and then all the real world product and business obstacles show up

@levelsio to clarify, we had booked almost nothing in advance biggest thing I felt we'd missed out on was the Ghibli Museum, which is booked out 3+ months (!!) in advance, even tho the tickets are like 5 euros

@levelsio I went to Japan with my girlfriend (who's Japanese) and we didn't have many issues. Japan does kinda have this booking-in-advance problem, but there's quite a few less known (sometimes even hidden) gems ime

Hakone seems fully booked months in advance, any alternatives for Ryokan Onsen style places? I'm going Tokyo then Osaka later so anywhere near there would be nice Apparently Japan is so popular you now have to book stuff 6 months in advance??????