Latent Space host swyx introduces the curation approach for the annual flagship conference AIE CODE, which is single-track, invitation-only, and aimed at about 1,000 agent engineers. This year's three themes are Scaling without Slop, System One models, and Increasing Ambition, among which Jarred Sumner will share Bun's rewrite in Rust and a Bun AOT preview. The article also rounds up that day's AI news, including Mistral's open-weight release and OpenAI's frontier model monitoring.
Shadaj Laddad, research lead for the AWS Hydro team, believes that code review can no longer keep up with the speed at which agents write code, and advocates placing correctness checks outside the agent trust boundary, enforced by the compiler—for example, the Rust borrow checker catches race conditions before runtime, while Hydro catches multiple classes of defects in distributed systems at compile time. The team is researching how to turn recurring design patterns such as idempotency into general-purpose guardrails, while acknowledging that compilers have limitations: AI-generated Rust code can compile successfully yet still be wrong. Hydro and the same team's open-source agent runtime Infinity are both jointly developed by AWS and UC Berkeley and hosted at hydro.run.
Shadaj Laddad, research lead for the AWS Hydro team, believes that code review can no longer keep up with the speed at which agents write code, and advocates placing correctness checks outside the agent trust boundary, enforced by the compiler—for example, the Rust borrow checker catches race conditions before runtime, while Hydro catches multiple classes of defects in distributed systems at compile time. The team is researching how to turn recurring design patterns such as idempotency into gene
Patrick, a product lead at Databricks, reviewed the cost overrun problem after the company fully adopted AI coding agents, saying that by migrating to cheaper models, routing, spend visibility, and cutting token overhead, unit costs in some scenarios fell by as much as 90%. He noted that hard budget caps would interrupt the best developers, so the company switched to gradual friction through warnings, approvals, and model downgrades, and productized its internal solution as Unity Gateway. The team built its own evaluation using real PRs from its own codebase, finding that GLM 5.2 matched Opus 4.8 in quality at about one-third lower cost per task, while a model that was 1.7 times cheaper per token ended up costing more per task because it consumed more tokens.
Patrick, a product lead at Databricks, reviewed the cost overrun problem after the company fully adopted AI coding agents, saying that by migrating to cheaper models, routing, spend visibility, and cutting token overhead, unit costs in some scenarios fell by as much as 90%. He noted that hard budget caps would interrupt the best developers, so the company switched to gradual friction through warnings, approvals, and model downgrades, and productized its internal solution as Unity Gateway. The te
Alex Heath asked Google CEO Pichai why Gemini 3.5 Pro ultimately was not released. Pichai said that while training the 3.5 series, the team found gaps in agentic capabilities, especially in coding, and believed that iterating with the smaller, more efficient Flash would be faster, so they released 3.5, 3.6, 3.7, and 3.8 Flash in succession; by the time this set of capabilities could be applied to Pro, Argon pretraining had already been completed.
Jellyfish's analysis of data from about 300,000 engineers at 1,300 companies shows that at the median enterprise, autonomous coding agents have already written 22% of merged PRs, up from 18% a month earlier, with the top 10% of companies reaching 53%. However, the median share of PRs independently reviewed and merged by agents is only 0.6%, and 9% for the top 10%, with human review remaining the main bottleneck. The author also notes that the abandonment rate for agent-submitted PRs is 40%, versus 20% for humans, and that the number of human reviews per merged agent PR has more than doubled over the past year.
Jellyfish's analysis of data from about 300,000 engineers at 1,300 companies shows that at the median enterprise, autonomous coding agents have already written 22% of merged PRs, up from 18% a month earlier, with the top 10% of companies reaching 53%. However, the median share of PRs independently reviewed and merged by agents is only 0.6%, and 9% for the top 10%, with human review remaining the main bottleneck. The author also notes that the abandonment rate for agent-submitted PRs is 40%, vers
The Guardian interviewed several San Francisco startups, describing the actual experience after AI agents took over tasks such as scheduling, coding, marketing, and recruiting as "colleagues." Anything has 15 human employees and about 150 agents; Lindy cut one-third of its staff and used agents to fill the roles; a McKinsey survey shows that about 40% of large organizations used AI agents in some form between May 2025 and June 2026. Employees report spending large amounts of time correcting agents' mistakes, and customers and job applicants are resistant to the way agents communicate.
The Guardian interviewed several San Francisco startups, describing the actual experience after AI agents took over tasks such as scheduling, coding, marketing, and recruiting as "colleagues." Anything has 15 human employees and about 150 agents; Lindy cut one-third of its staff and used agents to fill the roles; a McKinsey survey shows that about 40% of large organizations used AI agents in some form between May 2025 and June 2026. Employees report spending large amounts of time correcting agen
Nathan Lambert reviews the newly released Intelligence Index v5 from Artificial Analysis, which weights knowledge work, coding, science, and general each at 25%, and presents new model rankings. He notes that 8 closed-source models in the rankings lead the best open-source model, Kimi K3.1 Pro, and argues that the new index leans more toward measuring where frontier models are heading, potentially underestimating the role of open-source models in the economy. The article also discusses the pros and cons of shifting to private benchmarks, and how the business model of benchmark organizations charging model vendors to verify results affects credibility.
Nathan Lambert reviews the newly released Intelligence Index v5 from Artificial Analysis, which weights knowledge work, coding, science, and general each at 25%, and presents new model rankings. He notes that 8 closed-source models in the rankings lead the best open-source model, Kimi K3.1 Pro, and argues that the new index leans more toward measuring where frontier models are heading, potentially underestimating the role of open-source models in the economy. The article also discusses the pros
Tomasz Tunguz believes that, after coding and customer support, security will become the third product-market fit for AI. Citing a survey of 75 enterprise security practitioners, he breaks down security budgets excluding endpoints into five categories—security operations centers, SIEM, penetration testing, threat intelligence, and vulnerability management—and notes that each category has significant token consumption. The article also cites OpenAI President Greg Brockman's judgment that security spending should far exceed software spending.
Tomasz Tunguz believes that, after coding and customer support, security will become the third product-market fit for AI. Citing a survey of 75 enterprise security practitioners, he breaks down security budgets excluding endpoints into five categories—security operations centers, SIEM, penetration testing, threat intelligence, and vulnerability management—and notes that each category has significant token consumption. The article also cites OpenAI President Greg Brockman's judgment that security
Using four security incidents within the week of October 16 to 23 as its thread—including two PHP remote code execution vulnerabilities discovered by AI, the backdoored LiteLLM version v1.107.0 surviving on PyPI for about two hours and being downloaded roughly 49,000 times, the OpenCode console leaking third-party model keys for 1.4% of users, and Microsoft reporting that a human organization used the open-source multi-agent penetration framework ARTEX to attack the supply chain—the article discusses the common problems facing AI auditing. After examining seven papers one by one, the author points out that the faithfulness of latent reasoning self-decoding reaches at most 0.4788, sparse autoencoders are nearly blind to backdoor shifts, probes fail when targeted by the object being optimized, and all three mainstream audit tools originate from the audited party's ecosystem or require borrowing the audited party's permissions. The article also cites the NARCBench multi-agent collusion benchmark from Oxford and New York University, as well as the report by Anthropic and Theorem using Claude to write roughly one billion lines of Rocq formal proofs for 239 entry points across 123 open-source libraries.
Using four security incidents within the week of October 16 to 23 as its thread—including two PHP remote code execution vulnerabilities discovered by AI, the backdoored LiteLLM version v1.107.0 surviving on PyPI for about two hours and being downloaded roughly 49,000 times, the OpenCode console leaking third-party model keys for 1.4% of users, and Microsoft reporting that a human organization used the open-source multi-agent penetration framework ARTEX to attack the supply chain—the article disc
Hugging Face co-founder Thomas Wolf said that his AGI standard remains AI winning a Nobel Prize, but the AI researchers at Recursive quantized lm_head—which is usually advised to be kept at high precision—to 4 bits, and the result differs from bf16 by only 0.7 percentage points, which impressed him. He also mentioned that the model was released as open weights on Hugging Face, and congratulated Recursive. The quoted original Recursive post introduced that Flint is its first model, positioned as a fast local coding model, built by the automated research system Eureka Machine, reaching 125+ tokens/sec on a MacBook Pro and 150+ on RTX Spark, and built from scratch in under 4 months.
Hugging Face co-founder Thomas Wolf said that his AGI standard remains AI winning a Nobel Prize, but the AI researchers at Recursive quantized lm head—which is usually advised to be kept at high precision—to 4 bits, and the result differs from bf16 by only 0.7 percentage points, which impressed him. He also mentioned that the model was released as open weights on Hugging Face, and congratulated Recursive. The quoted original Recursive post introduced that Flint is its first model, positioned as
