AI news, October 29, 2026: Nvidia's open transcriber, Google Connected Apps, GitHub Agentic Workflows
Five stories from Wednesday, October 28, 2026 (US time) for solo makers who build with AI. Two are tools you can use or apply for now, one is a new automation option inside GitHub, and two show where big platforms are taking AI features for teams and everyday users. Figures are as the companies and outlets state them.
Nvidia releases Nemotron 3.5 Transcribe, an open-weight speech model for 40 languages
Nvidia published Nemotron 3.5 Transcribe, a promptable speech model that transcribes recordings in 40 languages, translates 31 of them to or from English and labels up to eight speakers in recordings up to an hour long. You choose the transcript style with a plain-language instruction, such as verbatim or cleaned-up written form, and can pass a list of expected names or terms. It uses about 4.3B active parameters, is released under the OpenMDW-1.1 license with BF16, FP8 and NVFP4 checkpoints, and Nvidia also shipped version 2 of its Granary speech dataset. It works on recorded audio, not live streams.
Why it matters for makers: meeting notes, subtitles and multilingual voice features can now start from a strong open model instead of a paid transcription API.
Sources: Hugging Face blog, Model card, TechCrunch
Google opens a Connected Apps Platform so developers can plug into Gemini and Search
Google launched a preview of its Connected Apps Platform, a single place to build integrations that users can call from the Gemini app, Google Search or both. Developers connect a cloud-hosted MCP server, or Android MCP through the AppFunctions APIs, then register, test and submit the app in the new Google Connected Console, which also shows traffic, latency and error rates after launch. Access opens in phases, starting with U.S. developers on a waitlist, and Google says more than 50 connected apps are already live.
Why it matters for makers: an MCP server you build once may become a way to reach users inside Google's assistant, much like SEO did for websites.
Sources: Google blog, TechCrunch, 9to5Google
GitHub Agentic Workflows is now generally available
GitHub Agentic Workflows, an open source project from GitHub Next, reached general availability at GitHub Universe. You describe a repository chore such as issue triage, CI failure analysis, docs updates or test writing in a Markdown file, and the tool compiles it into a normal GitHub Actions workflow powered by Copilot or another coding agent. Agents run read-only by default inside a sandboxed container, and a threat detection job checks proposed changes before they are applied. It is available on GitHub Free, Team and Enterprise; using the default Copilot CLI engine needs a Copilot license.
Why it matters for makers: a one-person repo can hand routine maintenance to an agent that runs on the same Actions setup it already uses.
Source: GitHub Changelog
Anthropic adds smart reports and pooled budgets to Claude Enterprise
Anthropic made smart reports, in-app surveys and pooled group budgets generally available for Claude Enterprise. A smart report samples a team's transcripts over a chosen period, groups the work into workstreams, and shows spend, deliverables and common friction points such as a missing connector or a failed tool call, along with suggestions like workflows to turn into skills. Pooled budgets give a group one shared monthly limit with alerts to admins at 50%, 75%, 95% and 100%. Anthropic says the reports are not meant for judging individual performance.
Why it matters for makers: the feature targets large teams, but it signals that measuring what AI spend actually produces is becoming a product category.
Sources: Claude, TechCrunch
Chrome's enhanced autofill can now pull details from Gmail
Chrome's enhanced autofill now uses Gemini with Personal Intelligence to suggest details found in Gmail when you fill in a web form, such as order numbers, package tracking numbers and flight or hotel bookings. Suggestions are marked as coming from Gemini, can point back to the source email, and the feature can be switched off in Chrome settings. It is rolling out in the U.S. on desktop, Android and iOS for Gemini in Chrome users who have Personal Intelligence turned on.
Why it matters for makers: forms on your own sites may increasingly be filled by the browser's AI, so clear field labels matter more.
Sources: Google blog, The Verge