In partnership with
A Single Typo Shouldn’t Cost You a Sale
A shopper types “running shos” and gets nothing back. Another types “red waterprof trail runners under $150” and hits the same wall. Both knew what they wanted.
Traditional site search relies on exact keyword matching, leaving shoppers stranded at “0 results found” screens.
Gladly understands what shoppers mean, typos and all. Instead of a dead end, they get products that fit what they asked for and a conversation that keeps them shopping.
Hey there, 3 minute read.
🚀 Big Story: Google’s best model is locked to defenders
Google announced Gemini 4 Carbon on Oct 28 and says it ranks #1 on the Artificial Analysis Intelligence Index (v5.0) with a score of 59. Almost nobody can use it yet.
Built for security work: Google reports 88.2% on SEC-bench Pro (finding vulnerabilities), 54.8% on ExploitGym (exploitation) and 71.0% on CWE-bench (patching).
Who gets it: a subset of cyber defenders now, widening through Google’s Fairwind Program. Developers, businesses and consumers have no date yet.
The price: $2 per million input tokens and $10 per million output at launch, rising to $4 and $20 after the introductory period.
The caveat: the “10x better patches” figure comes from Google’s own Chrome team, and the other internal numbers are Google’s own tests.
Source: Google, Oct 28
Jason’s take: The strongest model in the world is now one you may not be allowed to sign up for. Labs are handing their best releases to defenders first, so build workflows that survive a model swap instead of betting your stack on one name.
⚡ Quick Hits
Alphabet’s cloud revenue doubled. Q3 revenue was $127.6B (+25%), Google Cloud grew 103% to $30.7B, and the 2026 capex guide rose to $195B–$205B, with 2027 set to top $300B. SEC filing, via AI Weekly
Microsoft books about $10B a quarter from OpenAI. Its 10-Q puts OpenAI-related revenue at nearly 11% of a $92.3B quarter, with $7.2B in receivables. Azure grew 48%. TechCrunch
DeepSeek and Moonshot reportedly trained on GPT-6 Astra outputs. The Information says both used distillation; DeepSeek’s browser-task score rose from 70.9 to 75.0 at $0.22 per call versus Astra’s $10.77. The companies declined to comment. The Information, via AI Weekly
Computer science enrollment is falling fast. Preliminary fall 2026 data show CS enrollment down 12.4% at four-year schools, 12.3% at two-year schools and 10.4% in grad programs, while health majors grew. The report gives no cause; coverage links it to AI shrinking entry-level coding work. National Student Clearinghouse
Epic says AI now writes Unreal Engine code. The company confirmed it after CEO Tim Sweeney announced a “huge pivot” to AI acceleration on Oct 12, and its engine lead has named Claude Code and OpenAI’s Codex among the tools. GamingBolt
Claude Code’s status line may be eating your quota. A widely shared third-party gist measured a background “agent_summary” feature at about 16% of its author’s Claude Code spend and cites four other measurements of 18–26%. We found no Anthropic response, and the gist’s fix is an unofficial script, so treat it with care. Flamerix gist
💬 What people are arguing about
“Carbon is #1, but who can use it?” Arena ranked it first in Text Arena at launch, while several creators’ videos lead with the catch that you can’t try it yet.
Who trained on whom: the distillation report reopens the fight over whether cheap models are standing on US labs’ outputs.
Entry-level coding jobs: the enrollment drop feeds the argument over whether AI is thinning the pipeline.
Bailout talk: House progressives filed a nonbinding resolution against exceptional support for AI firms, citing Anthropic’s S-1 numbers. The American Prospect
📺 Trending on YouTube
🛠 The Workflow: find where your AI budget leaks in 10 minutes
Open your usage dashboards (OpenAI, Anthropic, Google, your coding tools) and sort the last 30 days by spend.
Name your top three workflows by cost, and note which ones run in the background without you asking.
Rerun one on a cheaper model with 10 real samples. The distillation report shows how wide the gap can get: $0.22 versus $10.77 per call on one browser benchmark.
Set a monthly cap and an alert on every key and seat, so one background job can’t drain the month.
Reply and tell me which workflow surprised you.
🧰 Trending Tools
Gemini 4 Argon API: Google’s $2 / $10 model, available to developers now. For builders who can’t wait for Carbon.
Claude Enterprise Smart Reports: analyzes what a team uses Claude for and what it costs, with 10 free reports a month. For admins and ops leads; an owner has to switch it on first.
Rime Prism: a speech-to-speech model with a text layer you can inspect or veto, and about 0.57s to first audio. For teams building voice agents.
Chrome autofill with Gemini: opt-in, fills forms using flights, orders and memberships found in your Gmail. For anyone tired of retyping; US desktop first, so check the privacy settings.
Hy-Nomas: an 8B open-weight interactive world model with a recipe for a single RTX 4090. For tinkerers exploring interactive video.
Tools are from AI Weekly’s roundup and vendor pages. Capability claims are the vendors’ own.
Analytics on Live Data Without Leaving Postgres
When analytics on Postgres slows down, most teams add a second database. Then they manage pipelines, sync lag, and drift forever. TimescaleDB extends Postgres instead. Analytics run on live data, in the database you already have. No pipeline. No migration. No new query language.
Tomorrow
Whether Gemini 4 Carbon opens beyond defenders, and how the labs answer the distillation report. Hit reply and tell me what you want covered.
