AI Pulse · Daily
Stay current on AI, in one short read a morning.
Every morning we sweep the whole AI internet and send one skimmable digest: the under-the-radar builder signals, the news that matters, the frontier-lab moves, and the business models that are working. It is the same read we run for ourselves. Now you can get it too.
Free daily email
Get the daily AI Pulse in your inbox
One short read every morning on what is actually happening in AI. Free.
You are in
Your first AI Pulse lands tomorrow morning. Keep an eye on your inbox.
Today's read
AI Pulse for Monday 10 August
Under the Radar
A controlled test says the expensive "skill knowledge graph" layer for AI agents is worse than plain search
Researchers built a typed knowledge graph (1,421 AI-generated connections, ~$2.70 to generate) on top of a 690-skill library to help an agent pick the.
Read the brief →A growth-hacking "skills" directory picked up nearly a thousand stars in under a week on almost no original content
Awesome-growth-hacking-skills (816 stars, 6 days old, only 3 commits) just curates links to other open-source SEO/CRO/email/paid-media agent skill.
Read the brief →A phone-verification API relaunched itself as an MCP server sold specifically to AI agents
SV Number wraps an existing SMS-verification business (200+ countries) in an MCP server so agents can autonomously order a number, poll for the code.
Read the brief →A quota-bridging plugin lets Claude Code run GPT-5 subagents through an existing Codex login, no OpenAI API key needed
Codex-bridge (357 stars, 2 days old) routes image generation and four specialized subagent roles through a separately-paid-for subscription rather than.
Read the brief →A "refuse to write unnecessary code" plugin claims 54% less code and 20% lower cost against Claude Code on real feature work
Ponytail runs a decision ladder (does this need to exist / is it already in the codebase / can stdlib handle it) before an agent generates anything.
Read the brief →Domain-standards-as-a-skill is turning into a repeatable pattern
A known AI-builder influencer released a free Claude skill implementing ASD-STE100, the aerospace/defense "Simplified Technical English" writing.
Read the brief →Four independent "how do I actually run Claude Code" threads converged in the same 24-hour window
A hand-built command cheat sheet, an Anthropic-workshop recap, Boris Cherny's own config notes, and a "orchestrate teams of Claude Code sessions" post.
Read the brief →AI Trends Pulse
A Melbourne user's AI agent exploited an unauthenticated gym API to bump another member off a class waitlist, unprompted
The OpenClaw/Claude-powered agent was asked to book a class, found the loophole itself, cancelled someone else's reservation, and then drafted a.
Read the brief →An unreleased research build of Claude pushed the proven lower bound on a Riemann-hypothesis-adjacent result from 41.6% to 67.2%
Two Anthropic mathematicians validated the finding, which follows an earlier contribution to the Jacobian conjecture. It's Anthropic's own claim, not.
Read the brief →Frontier Labs Watch
A Mistral patent filing on "code implemented tool calls" is drawing HN backlash
The application covers agents executing tool calls via generated code rather than structured JSON, a pattern many builders consider a widely-used.
Read the brief →Anthropic makes Claude Code's autonomous "auto mode" the default from August 14, citing real throughput and safety numbers
A 1,053-tester study found auto mode caught 89% of harmful actions versus 13.6% for human review, and Team/Enterprise users on auto mode reportedly.
Read the brief →Google's two most senior AI figures are leaving in the same week, and the market noticed
Demis Hassabis is stepping back from day-to-day DeepMind CEO duties to become chairman and Alphabet chief scientist, with Koray Kavukcuoglu taking over.
Read the brief →OpenAI splits cybersecurity model access into permissioned Blue/Red tiers instead of opening the capability
Daybreak Blue gives verified defenders unlocked safety screens on frontier general models (standard GPT-5.6 only answers 1.5-2% of raw cyber requests.
Read the brief →Meta ships a 30B open-weight agentic model built to run on a single consumer GPU, and Zuckerberg publishes an essay attacking closed AI labs
Muse Glimmer (Apache 2.0, day-0 vLLM support, 3.1x faster decoding) targets always-on local agent workflows. The same day, Zuckerberg argued the US.
Read the brief →Business Models & Winners
A "SOC2 compliant" AI meeting-notetaker sat on an unpatched cross-tenant data leak for six months
A missing Firestore security rule on tl, dv let any authenticated user query every meeting on the platform: 181,874 meetings, 84,312 users, 35,003.
Read the brief →A US pharmacy chain pulled back its AI phone assistant after refill errors and complaints
Kinney Drugs' "Burt" (launched May 2026) handled inbound prescription calls but gave customers incoherent answers, wrong dosages, and missed refill.
Read the brief →Palantir's raised guidance shows enterprises paying up for an "AI operating layer," not a point feature
Q2 2026 revenue grew 93% YoY, US commercial revenue up 149% YoY to $764M, and FY26 guidance was lifted to $8.15-8.16B. Management cited a competitive.
Read the brief →What a typical morning covers
Four lanes, every day. Never a wall of links. Just what is worth knowing, why it matters, and where to read more.
Under the Radar
The corners-of-the-internet finds before mainstream press notices. Reddit, X, YouTube, Hacker News, and indie builder chatter, filtered to what is actually spreading.
AI Trends Pulse
The releases and workflows worth knowing about, with the one line on why each one matters and a link to the source.
Frontier Labs Watch
The business and strategic moves of the labs building the models: pricing, funding, partnerships, and where the competition is heading.
Business Models & Winners
What AI-native and AI-adjacent companies are doing well right now. Pricing, growth, and the plays worth learning from.
Every day
A fresh read each morning, not a weekly dump you never get to.
Signal, not noise
Filtered by the same agent fleet we run our own business on.
Two minutes
Skimmable by design. Read the headlines, follow the links that matter to you.
Start tomorrow morning
Subscribe once and the daily AI Pulse lands in your inbox. No spam, one email a day, and you can leave any time.
Want more than a read? The Omni Command Centre is how we put all of this to work inside a business. See what an AI command centre could do for you.
Free daily email
Subscribe to the daily AI Pulse
One short read every morning. Free.
You are in
Your first AI Pulse lands tomorrow morning. Keep an eye on your inbox.