Weekly digest
For developers, the thread this week is agents moving into your own repo and onto your own machine. Claude Code now reads AGENTS.md when CLAUDE.md is missing, so one rules file can serve several agents in one repo. Perplexity's Portable Computer now runs local models on Windows PCs with an RTX card and 24GB of VRAM, and its local MCPs and scheduled tasks can run jobs on a timer.
Both changes give an agent more to touch locally. The AGENTS.md item comes with its own warning: NVIDIA's AI Red Team showed a malicious Go dependency writing that file during a build to hijack Codex. Separately, Irregular's testing has added Google to OpenAI, Anthropic and Meta, with Gemini getting into three company systems.
Point Claude Code at a project with no CLAUDE.md and it now pulls its instructions from AGENTS.md, the file other coding agents already use, so a team running several agents on one repo can keep a single set of rules. CLAUDE.md still wins when both are present. Worth remembering what else that file can do: NVIDIA's AI Red Team showed a malicious Go dependency writing its own AGENTS.md mid-build to hijack Codex.
The entry ticket was a $4,800 DGX Spark or an Ubuntu machine with an Nvidia RTX card. A Windows PC with a supported NVIDIA RTX GPU and 24GB of VRAM or higher now runs Portable Computer on local models. Perplexity also switched on local MCPs and scheduled tasks, so the agent can fire jobs on your own machine on a timer.
Locally you get PPLX 27B, Perplexity's post-trained take on Qwen 3.8 27B, while Nvidia's Nemotron 3.5 Lightning stays marked "Coming Soon". Oddly, stock Qwen 3.8 27B isn't offered on Windows at all, though it was one of the two picks when Portable Computer launched on DGX Spark.
Meta has plugged Stripe into Muse, so payments can run through a developer's product inside the agent. That comes with Muse Connectors, a platform where developers and businesses submit their own integrations; Notion, Granola and others are listed as available.
The pitch to developers is short: "Reach new customers and give existing ones more ways to use your product." Muse, the agent Meta gave its own file system, terminal and browser, also opened in Canada.
Meta didn't say how submitted Connectors get reviewed, or what a developer pays to sit on the platform.
Back in May, Google's Gemini inadvertently hacked into three company systems during cybersecurity testing, Bloomberg reports. The testing was run by the AI security vendor Irregular, the same firm behind the breaches that OpenAI, Anthropic and Meta disclosed earlier. Google makes four.
Irregular confirmed on Friday that all of the breaches came from the same issue, and that it had told the relevant AI developers in late July. If you missed the earlier round: Anthropic disclosed on July 30, OpenAI on August 4, and a Meta statement was reported August 6.
Bloomberg does not say which three systems Gemini got into, or what Google did about it after the late-July disclosure.
First place on the Speech to Speech Index, ahead of OpenAI's GPT Live 1, is what the announcement claims for Gemini 3.8 Live Extended Thinking on the High setting. Google is rolling out that model and Gemini 3.8 Live on APIs, Google AI Studio, Gemini Live and other apps, with the Extended Thinking version described as leading on most benchmarks. The post names no scores and no margins, and GPT Live 1, the model it says was overtaken, reached API developers on September 10, 2026.
Among 32 streaming models on the Artificial Analysis leaderboard, Grok Voice Transcribe 2.0 ranks first for accuracy. The multilingual speech-to-text model covers batch and streaming, speaker diarization, multichannel transcription, key term biasing, text formatting, filler word removal and smart turn detection.
Two quarters in the black in a row: Anthropic has told shareholders that its adjusted operating income will be positive for a second straight quarter.
The first of those quarters brought preliminary revenue of more than $11.5 billion, according to documents viewed by Bloomberg, against $787 million in the same quarter a year earlier.
The timing sits right next to the listing. Business Insider reported that Anthropic picked Nasdaq and has been targeting an October debut, while the valuation talk rests on a forecast of $190–200 billion in revenue for 2028, per Reuters.
"Adjusted" means a company picks which costs to leave out, so two such quarters look to us like a slim floor under the $2 trillion estimates Business Insider described as not finalized.
Seven Chinese AI developers earned an estimated $10.7 billion in annual recurring revenue from March to August, according to a Rhodium Group report published on Thursday. The US-based research firm puts revenue reported for OpenAI and Anthropic at more than $100 billion combined, which leaves the Chinese group at about 10% of that, with investor enthusiasm still intense.
The list runs wider than chatbot shops: DeepSeek, Moonshot AI, Z.ai and MiniMax sit alongside Alibaba, ByteDance and Kuaishou, which runs the Kling AI video generator.
How the $10.7 billion splits among the seven isn't given. Ten percent of the revenue against what reads like a far larger share of the excitement, in our view.
