Show It Once. Let ChatGPT Learn the Work.
OpenAI’s new Mac features move AI from answering instructions to learning from work history and demonstrated routines—but key control questions remain unanswered.
THE LOG — TRANSMISSIONS FROM ALTITUDE
What we learn running AI systems every day — workflows, growth loops, tools, and the occasional hard lesson. Written down so you can reuse it.
OpenAI’s new Mac features move AI from answering instructions to learning from work history and demonstrated routines—but key control questions remain unanswered.
Google says AI Studio Build can now start from an existing GitHub repository and push and pull changes across environments, although key workflow limits remain unexplained.
Claude says it can send Gmail replies and manage Google Drive files, with users controlling when approval is required.
Claude Code’s /design research preview adds an editable, human-controlled decision layer between a UI request and implementation.
Sakana Namazu brings Japanese-specialised reasoning to OpenRouter, but business use depends on training, privacy and data-residency checks.
HeyGen says longer avatar performances, code-driven motion control and two-host generation can bring more of a creator’s production workflow into one system.
OpenRouter says GPT-5.6 Sol is half off, but the provider route, key arrangement, tier and promotion window must align.
Z.ai says stronger post-training made GLM-5.3 a more capable coding agent while unexpectedly accelerating multi-stage cyber reasoning, prompting a safety-gated release.
OpenAI’s limited Ultrafast preview tests whether frontier intelligence can keep pace with live work, while its headline speeds remain supplier-reported maximums.
Claude says Chrome sessions can continue across desktop, web and mobile, bringing saved context and connected tools—but browser action expands the trust boundary.
Google says Gemini 3.7 Flash improves coding and agent workflows three weeks after 3.6, at half its original introductory token price—but supplies no benchmarks or absolute price.
DeepSeek’s developer-preview agent harness treats models, tools, sessions, sandboxes, orchestration and UI as replaceable plugins—but practical interoperability remains unproven.
xAI says Grok 4.6 can sustain ambitious work across research, building, self-testing and revision—but the evidence remains supplier-reported.
Claude in Chrome can continue across desktop, web and mobile, but smoother handoffs do not remove the risk of hidden instructions inside web pages.
Spotify’s Xirp proposes a shared organisational-memory layer around coding agents, connecting services, owners, dependencies, decisions and session knowledge through Spotify Portal.
OpenAI says ChatGPT Work and Codex can import selected work from other agents and keep it updated, with important limits around compatibility, fidelity and availability.
Google DeepMind says SL2T turns on-device pose landmarks into server-translated text, bringing ASL input to Gboard and Live Transcribe on Pixel 11 first.
Koray Kavukcuoglu takes operational command of Google DeepMind while Demis Hassabis shifts towards Alphabet’s longer-term AGI and science direction.
Kimi pitches Slides as a research-to-presentation workspace where people can inspect and revise the generated structure, charts and drafts before downloading the deck.
OpenAI is bringing ChatGPT, ChatGPT Work and Codex into supported Linux desktop workflows, but the first preview has a tightly defined support matrix.