Welcome back! Today’s AI stories follow a clear drift: systems moving from the sidelines into command rooms, onto our faces, inside our chats, and deep into lab notebooks. Defense planners lean on commercial models, product teams turn eyewear into screens, engineers let bots touch their code, and researchers ask agents to walk them through complicated biology. The tools keep slipping closer to the places where judgment lives, and today is a tour of how casually we’re starting to accept that.
In today’s Generative AI Newsletter:
Pentagon taps Google to staff AI war room.
Google spends $150M chasing Warby Parker glasses.
Anthropic wires Claude Code into Slack bugfixes.
SciSpace BioMed agent guides everyday lab questions.
Latest Developments
The Pentagon has placed Google’s Gemini for Government at the center of GenAI dot mil, a platform meant to guide three million people across the War Department. Officials describe the rollout as a leap toward smarter operations, yet the language reveals something larger. The U.S. is beginning to treat AI as a standing member of the national defense establishment. Leaders described the rollout as a foundational step in preparing the force for an AI shaped future and the scope hints at a rewrite of how decisions move through the system.
What the Pentagon Announced:
Scope of Adoption: GenAI dot mil will reach about three million workers across military and civilian branches.
Core Platform: Google’s Gemini for Government becomes the first model embedded in the system.
Official Messaging: Secretary Pete Hegseth said, “AI is America’s next Manifest Destiny,” describing the tool as essential to future warfare.
Contract Landscape: Google secured a two hundred million dollar agreement, with OpenAI, xAI and Anthropic also receiving major defense contracts.
The War Department calls this a strategic imperative. The world’s largest military institution is placing commercial AI at the center of its daily rhythm, blurring the line between public power and private platforms. Officials speak with the ease of a startup launch, even as they oversee immense authority. A senior aide described GenAI dot mil as a cultural shift. Cultures shape instincts, instincts shape judgment, and judgment is the last thing anyone should place on autopilot.
Special highlight from our network
AI voice tools are only as good as their ability to understand you. Speechmatics cuts real-time transcription errors by 25% compared to the next-best model. It performs across accents, noise, and overlapping voices.
That means more accurate, reliable conversations. Built for developers. Ready for enterprise.
Special highlight from our network
Not many people are talking about these “deep work” AI agents yet.
With Incredible, you just describe a workflow in plain English and it does the rest.
In 30 seconds, you get a working AI agent. No prompts, no wiring.
Example.
New lead comes in, the agent researches them, enriches your CRM and drafts a personal intro email—all on its own.
It runs on Agent MAX, their new agent model that
Holds up to 100x more context than typical AI
Delivers near zero hallucinations by checking actions against real data
Runs up to 90% cheaper than other agent platforms
Ready to see what an agent can do for your team?
Use code EARLYACCESS90 to get 90% off the Pro plan for 3 months
That is 75 credits for only $7.50 if you activate before December 19
Build your first agent with Incredible → Try Incredible Agents
After the failure of its early 'Google Glass' experiment, Google is coming back with two types of AI glasses and a wired XR pair built with XREAL. It's investing up to $150 million in a Warby Parker partnership to design AI glasses that look and work like regular eyewear. Google says the first glasses will arrive in 2026 with a promise to help with translation, navigation and memory. These glasses will have cameras and microphones positioned at eye level for enhanced functionality.
Here’s what sits behind that clean launch slide:
Hardware: Two AI variants (screen-free and in-lens display) plus wired Project Aura XR glasses.
Money: Up to $150M for Warby Parker intelligent eyewear, split between product development and potential equity.
Use cases: Gemini whispers directions, handles real-time translation, identifies objects and recalls moments from your day.
Concerns: Always-on sensors, cloud-based processing and tiny recording lights that many people may miss.
Apple plans to release its smart glasses in 2026, Samsung is promoting Galaxy XR, and other companies are striving to transform glasses into the next smartphone screen. If this bet works, 2026 could signify a shift from focusing on phones to embracing ambient AI by looking up casually. If it goes wrong, we may find that 'hands-free' convenience can turn into harmful surveillance seamlessly. The balance between convenience and potential invasion of privacy will be a key factor in the success of this technology.
Earlier this week, Anthropic wired Claude Code into Slack so bug threads can turn into pipelines. When someone tags @Claude under a messy report, Claude Code lifts the conversation, picks a connected repo and assembles fixes instead of suggestions. This integration streamlines the process of addressing bugs by automating the connection between bug reports and code repositories, without sifting through multiple platforms for information.
This is how a tag now unfolds:
Trigger: Tag @Claude and the bug thread becomes a coding session on your linked repos.
Session: Claude Code spins up a run, edits files, runs checks and prepares a draft pull request.
Updates: It drops progress notes in the same thread so teammates can steer it in plain language.
Guardrails: Sitting between Slack and your repos, it forces decisions about permissions, logging, and what it should never touch.
Other tools like Cursor and Copilot sneak into chat, but Claude Code leans harder on Slack as a command center for discussion of implementation inside Slack. On the bright side, teams get faster fixes and less context switching. On the darker side, a Slack outage, a misconfigured permission, or one overeager prompt can stall a release or leak secrets. That trade-off defines AI tooling: whoever owns the conversation owns the workflow, and you decide how much of your development you let live inside a chat app.
Special highlight from our network
Retail shoppers hate feeling rushed or tricked, but with the rapid adoption of agentic AI, they can feel unheard and often misunderstood.
Leading retailers focus their AI experiments on crucial interactions instead of pushing people down a funnel. Smart agents help compare options clearly, surface fees, return policies, and give staff enough context to have real conversations.
In a new session ‘Designing for Dignity,’ Thoughtworks explains ways to use agentic AI so that the customers feel heard, informed and in control of the choices they make. If you shape AI in retail, it helps you swap tricks for trust and build journeys that protect both revenue and dignity.
Sign-up here
SciSpace BioMed Agent helps you work through biomedical questions without writing code or managing analysis tools. You describe what you need, upload a file if you have one, and the agent builds a clear, step by step workflow. It is useful for students, new researchers, and anyone learning how to move from raw data to biological insight.
Core functions (and how to use them):
Gene and variant interpretation: Paste in a gene list or patient variants with a short note on symptoms. The agent points out which genes are most relevant and why.
Single cell support: Upload an .h5ad file and ask for cell type labels. The agent produces plots and summaries you can understand without knowing the underlying software.
Paper reading help: Ask a focused question like “what is known about pathway X in lung disease?” It returns a short explanation pulled from recent research.
Drug exploration: Give a disease mechanism and a few drugs you are curious about. The agent highlights which ones match the biology you described.
Basic workflow guidance: Request outlines of common experiments such as CRISPR screens or RNA seq pipelines. The agent explains the logic in plain language.
Try this yourself:
Pick one small task you already know, like understanding a gene you keep seeing in your project. Type the gene name and your question into the agent. Review the explanation and the papers it references. Use this as a learning exercise to see how experts normally structure the reasoning.
