Rowan Trollope

@antirez Patch edit provides the surrounding few lines with the crc32 of what’s being replaced. You could see how this could be quite problematic if the window is too small. If a variable is declared at the top of the function and the name of it changes for example.

@rowantrollope The problem with that is that even if the file got unrelated changes, the edit will fail. It's a big tradeoff.

New blog post: a new EDIT tool for LLM agents: antirez.com/news/166

@jerryjliu0 @mintlify Have a look at github.com/redis/agent-fi… We implemented file search (grep), file search (bm25) and file query (vector sim)

Real question: what is the actual latest state-of-the-art for file search and retrieval? - Actual grep over filesystem - Virtualized grep / BM25 over a db (what @mintlify did) - Vector search over a db - Hybrid search over a db - SQL - none of the above - some of the above?

Happy Monday. On today's show: - @MikeIsaac (NYT) - @JoannaStern (Author) - @rowantrollope (Redis) - @DLeitersdorf (Decart) See you on the stream.

Be sure to tune in to @tbpn live at 12:10pm PT to see @rowantrollope on the stream talking about Redis Iris, building with AI, and more.

@vivoplt @karpathy @fchollet @ylecun @AndrewYNg @rasbt @dair_ai @lilianweng @jeremyphoward @simonw @_akhaliq @ID_AA_Carmack @gwern @goodside @drfeifei @demishassabis @antirez

15 AI related accounts you should follow on Twitter: 1. @karpathy 2. @fchollet 3. @ylecun 4. @AndrewYNg 5 @rasbt 6. @dair_ai 7. @lilianweng 8. @jeremyphoward 9. @simonw 10. @_akhaliq 11. @ID_AA_Carmack 12. @gwern 13. @goodside 14 @drfeifei 15 @demishassabis Let me know who I missed guys

Three easy steps 1. Connect your data sources 2. Define your data schema 3. Expose to agent through CLI or MCP Then watch your context improve over time with Agent Memory Server - our state of the art memory platform

Announcing Redis Iris. The context layer agents have been missing. Redis is already the most used database for agent data Today we’re making it easier than ever with Iris. redis.io/iris/

@antirez 100% !!

I must admit that nothing about computers, since I'm in love with the field, was so uninteresting as the Javascript different fashions, waves, frameworks, rewrites, hypes. And I'm one that loves almost every shit programming related.

Same Codex rules

I am the best OpenAI Codex Advertising, even if I would not sponsor a single video in my YouTube channel. But I believe on saying everybody when I think something is outstanding. I do it with restaurants, wines, persons I like, and AI systems.

Gentle reminder on how, in the recent DS4 fiesta, not just me but every other contributor found GPT 5.5 able to help immensely and Opus completely useless.

Excited to be joining the @docusign board as they make the transition into Agentic Contract Workflows! The future is entirely Agentic driven automated contract and agreement management - Docusign has an amazing vision and leadership position to drive this

OpenAI winding down fine tuning is an interesting development and one to watch. On one hand, model maximalists will argue the largest models keep getting better at more things, so the need to adjust the weights of them is less necessary. On the other hand, the big labs keep pushing their models to a handful of use cases while training their harness designs into the model, rendering them less generalized. There's an argument _this is fine_, because coding and reasoning abilities will solve most other problems. But what we end up with are models build for their own harnesses. @badlogicgames was wrestling with Claude in the OSS Pi harness this week, trying to wrangle out specific in-harness behaviors, with Claude fighting him every step of the way. If this continues, there's a world where 3rd party harnesses become less valuable when used with frontier lab models because the 1st party harness behavior is already _baked in_. And there's no longer a fine tuning escape hatch to generalize this behavior away. Will then frontier models resemble appliances, not general platforms? With their harness trained in and no ability to adjust it? This might make application building easier for some enterprises, but the trade off is lock in.

Nice work Salvatore!

Yet another banger from @JayaGup10. The shape of a company's talent identity does indeed become the long-term moat in many services-type industries (e.g., McKinsey, Goldman). One thing this triggered for me is something I've been thinking about for a while - the concept of "status" (as Jaya mentions). If we rewind, there was a lot of discussion about a retreat from status and the "elites" in ~2016-2024. "College is dead." "Legacies are dead." "Stop chasing brand names." The book "Excellent Sheep." etc. As the child of Indian immigrants, I was raised 100% for status. I'm a status monster! My parents (I kid you not) had two posters above my bed from childhood: 1. Harvard 2. Einstein I certainly never became the latter, but after my mom moved out of my childhood home, it turned out she had saved Albert's picture and gave it to me! I still have the big man above me. My mom's concept of status was also always fixed in earlier times. To this day I hear "you could still become a doctor - it's not too late." Of course, I still believe doctors are so important. But along the way, after every job change, mom would say "maybe you can work at IBM now." [no disrespect to IBM - iconic company but def from a different era] The point is status is "sticky." My New York friends always had status on the top of their minds. Goldman, Morgan Stanley, JPMC, Bain, BCG, McKinsey - duke it out for who is the best! Over the years for me, in starting companies, status became more about the company's status, which waxed and waned during each year. I am quite happy to have the status of an exited unicorn though. 😂 In COVID, it felt like status got quieter again, with everyone working from home. You just didn't get that feeling of constant in-person comparison at bars or events. Now, as I spend time with founders and am thinking about starting my own thing, I feel people checking their status HARDCORE. People are incredibly successful at 22. They all know each other. Because almost everyone in high growth tech is in SF, the comparison is constant at every house party or trivia night. The status rises and falls with each model release. OpenAI v Anthropic. Harvey v Legora. Sierra v Decagon. Which VC is the one to get status from? Thrive is the hottest new show in town, right? In the world of 10 years ago, status felt durable - e.g., McKinsey and Goldman have mattered for most of last many decades. But now, status is as ephemeral as benchmarks. What does it mean when young people hitch their self-worth to status that could turn on a dime? I think this is an interesting question to ponder. What I've learned over many years is human psychology is much more fragile than technology.

Cool to see how OpenAI uses Redis to scale voice AI to 900M users: openai.com/index/deliveri… We're huge Codex fans here at Redis - I use Codex 5.5 extensively - amazing work by the team there. GPT 5.5 is the best model for most tasks IMO. It even fixed my wifi tonight (no joke)

I just submitted a PR for a new Redis Array type :) github.com/redis/redis/pu…