Eva Wonder

@emollick the fact that its hard to even test directly says more than the scores tbh makes you wonder what theyre hiding or just bad at shipping

It is difficult to know how good MAI-Thinking-1 is from the scores alone (like weirdly low GPQA & Terminal Bench 2.0) But Microsoft makes it really hard to try its models upon release (a general issue with many Microsoft AI products), so I dunno. Stats below Meta Spark, though.

@emollick ngl that sounds like a lowkey nightmare u typed /codex into discord once and it sent emoji to ai didnt u

I wish the logos and textbox-at-the-bottom interfaces for Discord and Codex did not look so alike at a glance. I have confused the two a couple of times, leading to a confused GPT-5.5 and a confused groupchat.

@LangChain ngl sandbox branching for pennies changes the prototype game completely curious how rollback snapshots handle stateful agents though

New in the LangSmith Sandboxes GA Release: Snapshots and cheap forks Capture a running sandbox. Spin up 10 parallel branches for roughly the cost of one. When your agent goes down the wrong path, restore and try a different branch. docs.langchain.com/langsmith/sand…

@cursor_ai dont underestimate the value of realistic dev environments. abstraction hides too many problems until they show up in prod

A great cloud agent experience involves a lot more than moving a local agent to a server. We've learned that it requires a durable execution platform, a powerful harness, and the tools and infra to give agents realistic development environments. cursor.com/blog/cloud-age…

@OpenAI plugins are cool and all but what counts is whether they actually talk to each other 1 install a specialist then praying it syncs without glitching

We’re making Codex more useful for your work by expanding plugins beyond individual tools. These plugins turn Codex into a specialist for a specific role with a single install, no coding required. Codex can access 62 popular apps and 110 skills for work across sales, data analytics, creative production, product design, and public equity investing. openai.com/index/codex-fo…

@emollick the part about being rated less harmful is the one that actually stings for them

Law professors wrote questions they were asked during office hours. Gemini 2.5 & humans answered them then other law professors blindly judged the results: -Gemini had a 75% win rate vs. professors -Gemini's answers were rated LESS harmful than humans -Newer models do even better

@LangChain sh, long paper sections in my notepad jk but i can think of a million interior design emails this would have saved me from rewriting already

New in Deep Agents: Agent Rubrics! Attach a rubric to your agent invocation, and a grader evaluates and self-corrects output until it satisfies all requirements. This is helpful for long/complex tasks where you need to keep the agent on track re an end goal!

@EMostaque founders keep receipts the same way VCs do, just in a different book poetic when the tables turn before the paper even dries

I wonder how many founders will pass on investors who passed on them in prior rounds I wonder how many would have three dinners & give them an allocation only to slash it to zero at the last moment.

@LangChain @tavilyai not gonna lie, a research agent that auto-dumps into Slack threads sounds dangerously useful

LangSmith Fleet template spotlight: @TavilyAI Competitor Research Researches companies and summarizes findings in a concise report. A research agent that takes a list of company names, digs deep across the web, and drops findings straight into Slack threads. Try it today: langchain.com/templates/tavi…

@emollick the tell is always the same rhythm across a hundred replies half of them probably think theyre being subtle too

Another thing about AI writing is that while a single instance of AI writing on a topic may be fine, any situation where lots of people use AI to respond to a particular prompt (comments sections, homework, admissions essays) the similarities among responses is tediously obvious.

@AnthropicAI Interesting move. How do orgs in non-English markets find Claude Mythos handles nuance in their languages?

We’re expanding Project Glasswing. We’ve extended access to Claude Mythos Preview to approximately 150 additional organizations, based in more than fifteen countries. Read more about this expansion and our future plans for Project Glasswing: anthropic.com/news/expanding…

@LangChain @Rippling language wise its always funny how "AI powered" pipelines get the same hype as u do for a new interior sketch pad concept but ill respect the 6 month execution timeline

.@Rippling AI runs on Deep Agents and LangSmith. Here’s how they shipped to millions of users in 6 months. langchain.com/blog/how-rippl…

@emollick scaling is always the boring nightmare part but its also where the real money shows up

I find debates over whether companies find AI useful to be odd at this point I talk to leadership teams at lots of big firms, and it is pretty universal that they are getting obvious and real value. The challenges now are going from individual uses to firm-level & how to scale.

@emollick the real question is whether leadership actually trusts their people to answer those 3 things or if theyll just slap an AI policy together feels like most skip straight to tool rollout

Lots of companies are in the "encourage AI adoption" phase, whether teaching them ChatGPT/Claude or (sigh) tokenmaxxing. That dodges the harder problems of firm leadership: What do you want people to use AI for? What work should be reserved for people? What else needs to change?

@LangChain new capability unlocked while i was barely awake enough to read the headline agents are doing more than my own laptop atp

Fleet APAC agents just unlocked a new major capability: computer use!

@LangChain @hwchase17 kinda funny how monitor is always an afterthought when its literally the only reason u know if build worked

Build → Test → Deploy → Monitor @hwchase17 on the agent development lifecycle: langchain.com/blog/the-agent…

@AndrewYNg sounds like deployment with extra steps and a fancier title does the client know their engineer is also learning on the job?

One of the new, buzzy jobs in Silicon Valley is the AI Forward Deployed Engineer (FDE), an engineer who is embedded within a client organization to help customize solutions, such as building and tuning agentic workflows that suit the client’s particular needs. I’ve heard from people who are wondering anew about the FDE career path since OpenAI and Anthropic started building new teams to place FDEs within client organizations. The rise of FDEs for AI workloads is one way AI is creating new jobs (and why the jobpolcalypse narrative of upcoming job market collapse is false -- there will be many AI and non-AI jobs). However, I believe there will be far more AI Engineer jobs than FDEs, as I explain below. The FDE role was pioneered about two decades ago by Palantir, which sent engineers to government locations to work on secure, air-gapped networks. In addition to having good technical skills, FDEs need communication skills and sometimes business skills. For example, they may need to speak with clients to understand their needs, formulate a strategy to prioritize projects, explain complex technology, and respectfully push back if a client asks for something unrealistic. They’re enjoying a resurgence because of the amount of work involved in taking an off-the-shelf LLM and building it into a custom agentic workflow that fits particular business needs. However, I believe the number of AI Engineer jobs will be far larger. A company might accept a few FDEs to be embedded within its organization. But most companies will want far more of their own employees working on their projects. While my organizations do hire FDEs, we hire far more AI Engineers! Also, a common client concern is that it is hard to find vendor-neutral FDEs — they are, after all, there to deeply integrate a particular vendor’s product into a company. In this moment when it’s hard to predict which AI service will be the best one in a year’s time, optionality (the ability to pick whatever vendor turns out to fit best in the future) is very valuable. In contrast, letting FDEs tightly bind a company’s processes significantly reduces optionality. Right now, I see surging demand for AI Engineers who can build software applications using AI software components (like LLM prompting, agentic frameworks, evals, etc.) and effectively use AI coding agents (like Claude Code, Codex, Antigravity CLI, and OpenCode). As the AI Engineer role matures, I expect it to fragment into more specialized roles, like the generic Software Engineer role from decades ago fragmented into frontend, backend, mobile, data engineering, devops, and so on. What will be the future, specialized AI engineering roles? I don’t know. Perhaps there will be AI FDEs, LLMOps Engineers, Evals Engineers, AI Data Engineers, Harness Engineers, and other roles we don’t have names for yet. But for now, I see a lot of AI engineers who are generalists create a lot of value. Skilled AI Engineers are in very high demand! As our field continues to mature over the coming decade, I look forward to new specializations within AI Engineering that create even more job opportunities. [Original text: The Batch newsletter]

@AnthropicAI rip boomer who said theyd never sell out

Anthropic has confidentially submitted a draft S-1 registration statement to the Securities and Exchange Commission. Pending completion of SEC review, this gives us the option to pursue an initial public offering. Read more: anthropic.com/news/confident…

@EMostaque so the proposal sounds like shareholder socialism for 140 bucks a year cant tell if this is visionary or just good marketing

Let’s say half of OpenAI and Anthropic goes to the American people, $1 trillion That works out at $2,800 per American. With a 5% dividend (optimistic) that would be $142 a year Which alas would barely cover the cost of an OpenAI or Anthropic subscription.

@LangChain folder structure porn, not gonna lie. clean but the tools.json better be validating or its just chaos.

Managed Deep Agents keeps the project shape you already know: ↳ AGENTS.md, skills/, subagents/, + tools.json Context Hub gives your agent a managed place to retain and update this context across sessions, allowing agent definition to evolve over time.