Mike Taylor
@hammer_mt ·
Context Engineering with DSPy https://x.com/i/broadcasts/1qKVmydomYnxB
Back
AI posts
AI posts
People and companies shaping AI. Filter your watched accounts, catch up on recent posts, or find the most liked and viewed.
Compiled by the Toolstash team · Updated
209 watched accounts · 2984 matching posts · Showing up to 3,000 recent stored posts. Counts are snapshots, not live X totals. This page refreshes every minute while visible. Last collected .
Every 📧@every ·
Context Engineering With DSPy https://x.com/i/broadcasts/1oKMvNPOvyRGQ
Santiago@svpino ·
A new agent just launched.
The mission is clear here: learn to do something specific, then repeat it over and over.
1. You describe the job you want the agent to do. 2. Give it context and access to tools 3. The agent goes and finishes the job 4. You can review and ask for changes 5. When you are done, you can set the job to repeat
Séb Krier@sebkrier ·
If the efficient markets hypothesis applied to AI discourse, more people would realize how wrong the singletonians were (despite being right about capabilities improving) and how prescient the multi-agent/CAIS frame was. Few remember early claims of unitary 'seed AI' type agents.
Tibo@thsottiaux ·
Day 25/
We are doubling all usage limits for all Pro 100/200/500 tiers and Business Premium, effective today.
We heard your kind feedback and won’t be reducing usage limits for Pro 200, so your Work and Codex usage will stay the same. We’re also going to continue offering extended limits for dots for now, and usage of your dot will still continue to be separate from your default usage.
And we'll of course also keep shipping better models and other efficiency improvements that increase your effective usage over time.
Miles Brundage@Miles_Brundage ·
Bay Area AGI pilled people can only think about one book or book series at one time
There was the Making of the Atomic Bomb era, the Three Body Problem series era, the Seeing Like a State era, the Caro/LBJ era….
Maybe we won’t get another era, just more essays and podcasts
merve@mervenoyann ·
hey, Le Chonk by @MistralAI is now on Hub! 🐱
to celebrate, we ship day-0 support in transformers and llama.cpp, as well as quants 🤗
read our blog for model deep dive, integrations and more! https://huggingface.co/blog/mistral-large4
Artificial Analysis@ArtificialAnlys ·
We’ve announced the Artificial Analysis Progress Index: a single capability scale for every language model since 2022, launched alongside Intelligence Index v5
The Intelligence Index is still our primary measure of current model intelligence. It uses the evaluations most relevant to developers today, so its mix changes as evaluations are added and retired, and scores from different versions can’t be compared.
The Progress Index is built for comparisons across time. It uses item response theory, a statistical method used to score standardized tests, to estimate one general capability score for each model from the underlying tasks in all our evaluations, including those that are retired or excluded from the Intelligence Index.
Use the Intelligence Index to choose a model today. Use the Progress Index to see how capability has changed over time.
The Progress Index builds on prior work placing models on a single capability scale, in particular @EpochAIResearch's Epoch Capabilities Index and the capability scale @AnthropicAI published in its Claude Mythos Preview system card.
Key takeaways:
➤ Since reasoning models arrived, the frontier has risen ~65 to 70 points a year, about double the 2023 pace: the highest score among released models rose 35 points in 2023 (GPT-3.5 Turbo 125 to GPT-4 Turbo 160), 67 in 2024 (to o1 at 228), 68 in 2025 (to GPT-5.2 at 296) and 56 so far in 2026 (to Gemini 4 Carbon at 352).
➤ Open weights models generally trail proprietary models by about 4 months: since January 2025, each new highest open weights score came 2-8 months (median ~4) after a proprietary model first reached it. DeepSeek V4.1 Flash 1017 is the highest-scoring model with released weights at 324, a level Claude Fable 5 reached in June 2026. Kimi K3.1 Pro (weights not yet released) scores 341, about two months after the closed frontier
➤ New frontier scores are being set more often: the highest score among released models was raised on 2 release dates in 2023, 4 in 2024, 9 in 2025 and 11 so far in 2026. That is about once every four weeks this year.
➤ OpenAI has held the top score most of the time since November 2022, but the frontier has become a close contest: OpenAI held the number one position for the first 27 months measured, but in 2026 especially there have been rapid changes in the leader between OpenAI, Anthropic, and now Google
Mia@MiaAI_lab ·
More of RTX PRO 6000.
The cards are BEAUTIFUL.
- Mia
@MiaAI_lab ·
My coding agents have a problem.
I personally have around 20 agents working for me around the clock. They're all fast, but the problem is not speed - it's DIRECTION!
When I change anything in the plan, an agent that doesn't know about it keeps building on the old one.
That wastes tokens, but more importantly - TIME.
Serro has a tool called Agentic TPM that tells ALL your agents when something in the plan changes. The agents get updated, but you still have the final word.
See how it works 👇
@jakejooyoung
corbin@corbin_braun ·
for every good trade
you need your canary in the coal mine
kache@yacineMTB ·
Bro please don't release allignment papers lmfao there are two sides to that coin
Sarah Wolf@sarahzorah ·
Scenes from Claude Founder House SF
Latent.Space: AI News & Views@latentspacepod ·
How does @ElorianAI's new "visual reasoning" model Glimpse compare to frontier models in vision tests? A new benchmark called DecoyCount provides some answers.
We quizzed CEO @AndrewDai on the benchmark, and also talked about Elorian’s positioning in a crowded models landscape. https://latent.space/p/elorian/
Arena.ai@arena ·
Lower cost per task doesn’t always come from lower token prices.
On Agent Arena, GPT-6.1 Sol (Max) by @OpenAI costs $1.91 per task: - 85% less than #1 Claude Opus 5.5 (Max) by @AnthropicAI at $12.51 - 87% less than #3 GPT-6 Astra (Max) by @OpenAI, at $14.93
But these savings come from different places.
Against Astra, Sol’s cost advantage comes primarily from lower token prices. Against Opus, it comes primarily from using fewer tokens per task.
How do we derive this?
Agent Arena tracks real costs of users’ tasks. Cost per task is the tokens a model uses times the price of each token, across input, output and cache-read tokens. By comparing pricing and token counts, we can identify what drives the cost reductions.
- Santiago
@svpino ·
This is for engineering and project managers.
It's a platform for coordinating engineering work across a team. The goal is to have a central place for people and agents to work together:
• It gathers info from connected tools (Slack, GitHub, Linear, etc.) • It organizes it by program • It maintains shared context about decisions, owners, risks, and progress • It lets you ask questions about what changed and why • It generates recurring updates and turns meeting outcomes into actions
Google Gemini@GeminiApp ·
You can now generate, preview, and edit documents, PDFs, presentations, spreadsheets, and more directly in Gemini.
Easily turn your handwritten notes into a PDF study guide, manually tweak content, highlight sections for Gemini to edit text or visuals, and even ask follow-ups.
- kache
@yacineMTB ·
holy shit Canada has an AI lab too
- kache
@yacineMTB ·
I think thinky machines are onto something
Thor 雷神 ⚡️@thorwebdev ·
I designed a custom voice for my ESP32 and have him tell me jokes. I call him Paddy and he's hilarious. 🤣
Ethan Mollick@emollick ·
I think “sell picks and shovels” is the wrong lesson for AI startups.
The Gold Rush analogy is weaker than the Industrial Revolution. Gold is a single pile that runs out. Machines keep making everything cheaper. Far more people got rich using the new machines than building them.
elvis@omarsar0 ·
Keeping coding agents aligned gets harder as you scale them.
They drift rapidly. Throwing more intelligence at it won't solve anything either.
Really like the direction @serroai is taking here. It captures plan changes, checks with the responsible person, and updates the relevant agents.
As we delegate more work, keeping everyone and agents efficiently coordinating becomes a much bigger engineering problem.
Kieran Klaassen@kieranklaassen ·
This is what I've been building since 2024
- kache
@yacineMTB ·
One thing I like about Carney is how he pretty much refuses to define Canada relative to America. He defines Canada as Canada, and answers the questions about how to deal with USA using answers about how Canada should be, irrespective to USA. bro actually read George grant?
- Santiago
@svpino ·
These guys built an agent that takes a process description and turns it into a TypeScript workflow.
Here is how it works:
1. You explain the rules you want to automate to the agent 2. The agent creates the automation workflow 3. You review the workflow and ask for any changes 4. When done, you can publish the workflow
After it is public, anyone at your company can run the workflow from Slack or Microsoft Teams.
Vox@Voxyz_ai ·
Change the plan mid-task, and unless your coordinator agent runs a top model, it 𝘄𝗼𝗻'𝘁 𝗲𝘃𝗲𝗻 𝘁𝗵𝗶𝗻𝗸 𝘁𝗼 𝘁𝗲𝗹𝗹 the agents working in the background.
They grind through the old plan and hand back "done" with a straight face. You burn tokens once for nothing and again on the redo.
On a team it's worse. Plans often change in meetings, and however smart your coordinator agent is, it 𝗻𝗲𝘃𝗲𝗿 𝗴𝗲𝘁𝘀 𝗶𝗻𝘃𝗶𝘁𝗲𝗱.
@jakejooyoung launched Serro today. It's an Agentic TPM, basically an AI version of a technical program manager. When the plan changes, it checks with whoever owns it first, then updates every person and every agent.
If it's just you, send this to your coordinator agent after you change the plan 👇
"I just changed the plan: [what changed]. Act as the program manager for every agent working on this: 1. Update the plan file first (create PLAN.md if there isn't one). If any part of the change is unclear, ask me before you pass it on. 2. Tell every agent that's still running what changed. Ask what it already built on the old plan, and stop anything that no longer fits. 3. From now on, have every agent reread the plan file before it starts and before it reports back. If the plan changes mid-task, it stops and asks first. 4. Every report covers what was done, where it is, how it was verified, and what's still open. 5. No agent pushes, deploys, or touches production without my OK. Then show me what needs to be redone."
Nathan Lambert@natolambert ·
This is very needed. Talking to models like Claude about anything that has an overlap with an area like legal, healthcare, etc is mind numbing. They’re so desperate to hedge and steer away from grey area that they often are really avoiding the real question.
Aravind Srinivas@AravSrinivas ·
/teardown by @jakubkrehel now available as a skill in Perplexity Computer. Allows you to break apart any thing and look into it in 3d interactively. 5000 trial credits per user to tinker around.
jack friks@jackfriks ·
so like 2 months ago i setup a grok bot to use @postbridge_ to mirror any post i make on here (X) to all other platforms
then i go look every few weeks and see i have posts that got a bunch of traffic for essentially 0 extra effort and a 1 time setup
hard to beat that
Marcus Williams@Marcus_J_W ·
Some details on what our monitoring system currently looks like https://openai.com/index/how-we-detect-frontier-model-misalignment/