# forck.live — 2–8 November 2026 > AI lab announcements, news reports and podcasts. forck.live follows what companies announce about their own products. Press may corroborate an official story; it never leads and never appears as an item of its own. A missing date stays missing. A title is the publisher's words — where a company does not publish in English, an English rendering is listed separately as `Title in English` and named as this site's translation, never as theirs. This file is the week in progress: every confirmed first-party announcement of ISO week 2026-W45, 2–8 November 2026, newest first. The window never moves, so this URL always names the same seven days — but they have not all happened yet, so the file is still filling: announcements are appended as they land, and it will answer with more of them later today. Fetch it again rather than reading a cached copy as the finished week. Every entry is one first-party announcement, and carries the same eight fields in this order before any others: Company · Announced · Category · Coverage · Announcement · Group · Source · Record `Source` is the company's own page the announcement was traced to and `Record` is this site's page for it. Both are safe to cite. `Coverage` is a number of outlets, or `not counted` — which means this record reached no count of its own, never that nobody wrote about it. `Category` is this tracker's own analysis, and is authoritative wherever it is stated. `not stated` means either that nothing has analysed the post yet or that the analysis placed it outside this taxonomy; neither is a verdict about the announcement. `Announcement` is whether the post is news of something the company did, rather than a how-to, a customer story, a changelog listing or a staffing note. `Group` is which of this record's groups the entry falls in, carried on the entry itself so it holds where the groups are headings and where they are not. One of: `models`, `covered`, `announcements`, `routine`. Anything after `Record` is held detail that only some announcements carry. - This week's page: https://forck.live/week/2026-W45 - The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md - Catalogue: https://forck.live/llms.txt ## Announcements ### Microsoft — Nine engineers, 1,211 pull requests in a day: Lessons from building Copilot Cowork - Company: Microsoft (microsoft.com) - Announced: 2026-11-02T15:13:56+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://commandline.microsoft.com/how-we-built-copilot-cowork-python-to-rust-migration/ - Record: https://forck.live/items/30555-nine-engineers-1-211-pull-requests-in-a-day-lessons-from-building-copilot-cowork - Subject: Foundry / Responsible AI In January of this year, a small team at Microsoft set out to build Copilot Cowork , an agentic execution layer that performs multi-step workflows on a user’s behalf. ### ElevenLabs — Introducing SFX v3 for developers - Company: ElevenLabs (elevenlabs.io) - Announced: 2026-11-02T12:00:00+00:00 - Category: new-model - Coverage: not counted - Announcement: yes - Group: models - Source: https://elevenlabs.io/blog/sfx-v3-for-developers - Record: https://forck.live/items/30561-introducing-sfx-v3-for-developers - Subject: Voice AI ElevenLabs is releasing SFX v3, a new sound effects model in its Sound Effects API that follows prompts with multiple instructions more closely than SFX v2 and handles vocals and music better. Developers switch by setting model_id to eleven_text_to_sound_v3 on the POST /v1/sound-generation endpoint, and the model picks the length of the sound effect from the prompt. If outputs drift, raising prompt_influence above its default of 0.3 makes generations follow the prompt more closely and vary less between runs. Looping still only works with eleven_text_to_sound_v2, so requests that need it should stay on v2 with loop: true. ElevenLabs calls SFX v3 an incremental update that is less likely to clip on high-frequency sounds like gongs, chimes and kettle whistles, and says its engineering team finds it sounds more natural overall. Because both models run on the same endpoint, the company suggests running common prompts through v2 and v3 to compare before switching. ### Hume — Introducing Hume's Audio Decision API: A decision layer for speech and sound - Company: Hume (hume.ai) - Announced: 2026-11-02 - Category: api-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://www.hume.ai/blog/introducing-hume-audio-decision-api - Record: https://forck.live/items/30566-introducing-hume-s-audio-decision-api-a-decision-layer-for-speech-and-sound - Subject: EVI / Octave Hume is introducing an Audio Decision API that it describes as a decision layer for speech and sound, aimed at contact centers that score every call against a rubric. Hume says changing what gets scored in a call is a project involving a new rubric, pipeline and evaluations, taking days if not weeks before the first answer. The transcript such scoring runs on captures only the words, according to Hume, missing the pause, the sigh and two people talking over each other. Hume says that information sits in the audio, in call recordings or ignored on real-time calls, and that until now there was no way to extract it in a form you can act on. ### Apple — Online Dense Reinforcement Learning with Planning-Grounding Agent for UI Navigation on Live iOS Apps - Company: Apple (apple.com) - Announced: 2026-11-02 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://machinelearning.apple.com/research/planning-grounding-ui-navigation - Record: https://forck.live/items/30541-online-dense-reinforcement-learning-with-planning-grounding-agent-for-ui - Subject: Machine Learning Research Apple researchers propose an online reinforcement learning framework that trains UI navigation agents directly on live iOS applications, arguing that frontier vision-language models are too costly for everyday use while smaller SFT or offline RL models memorize static trajectories and fail to recover from errors. The framework splits the agent into a Planning agent and a UI Grounding model, and uses an LLM-as-a-Judge to supply dense step-level feedback alongside a terminal reward for overall task completion. The agent is optimized with Proximal Policy Optimization and step-level advantage estimation, plus GiGPPO, a neighbor-conditioned value estimation method the authors say improves critic stability by drawing reward signals from visually similar states across concurrent trajectories. Supporting this at scale required infrastructure that orchestrates hundreds of concurrent virtual iOS devices. On UINavBench, iOSWorld and AndroidWorld, the online RL agents substantially outperform SFT and terminal-only RL baselines and converge faster, with the 8B agents showing nearly a sevenfold improvement over the SFT baseline in out-of-distribution generalization to Android and outperforming Gemini-3.5-Flash while tying Gemini-3.1-Pro on real iOS applications. ### Legora — Legora and Conga integrate to bring enterprise contracts into the agentic operating system for legal work - Company: Legora (legora.com) - Announced: 2026-11-02 - Category: partnership-acquisition - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://legora.com/newsroom/legora-and-conga-integrate-to-bring-enterprise-contracts-into-the-agentic-operating-system-for-legal-work - Record: https://forck.live/items/30528-legora-and-conga-integrate-to-bring-enterprise-contracts-into-the-agentic - Subject: Legora Legora and Conga announced an integration that brings contracts managed in Conga Contract Lifecycle Management directly into the Legora platform for in-house legal teams that use both. The integration is meant to cut the handoffs that force lawyers to download files, re-upload them elsewhere and reconcile versions by hand around contracts their CLM already governs. In the first phase, mutual customers can pull Conga contracts into Legora as templates for drafting or for analysis, legal research, summaries, redlining, editing, due diligence and Tabular Reviews. Conga's access controls carry through automatically, so users can only pull in contracts they can already access in Conga. A second phase will let customers save edits made in Legora back to the document's original location in Conga as a new version, a new document, or in place of the original, with access controls applied automatically. A third phase will let customers find contracts in Conga by describing them, such as every supplier agreement signed with a given vendor in the past six months. Legora says the integration joins a growing ecosystem of connections to systems enterprises already depend on, and that further detail and availability for each phase will be communicated to mutual customers over time.