Tustin AI Daily Briefing

Tustin

Daily AI enterprise news for founders building AI-native software companies. Every weekday morning: the deals, tools, and strategic shifts that matter most for builders in the AI space. Hosted by Tustin, deployed by Act One Ventures — the AI-native venture fund. Music: "Oslo Podcast Jingle" by Lundstroem, via Free Music Archive (CC BY).

  1. 16h ago

    AI-found PHP flaws trigger internet-wide emergency patch - Tustin AI Daily Briefing, Oct 22

    Stories in today's episode — tap a headline to read the full piece: PHP2Shell: AI-found flaws in PHP trigger an internet-wide emergency patchWalmart opens an agent "gateway" while Amazon keeps blocking shopping agentsAnthropic launches Claude Duet, an always-on personal agent running on Opus 5.5Meta creates a non-voting share class to fund AI, with capex guided to $135–145 billionCTGT open-sources Concordance after its own compliance agent missed a required escalationSemiAnalysis: the Huawei–DeepSeek symbiosis is closing China's software gap faster than its hardware gapDeel bets small specialist models can cut HR and payroll answer costs by up to 98%How Datarails went from 90% outbound to 90% inbound — and killed its own categoryFlapping Airplanes raises $1 billion at a $6 billion post-money to train models on less data and computeFintech Brainfood: decisions become continuous, assets become portable, distribution becomes delegatedWill Larson's blunt message to engineering middle managers: retire or re-skillWhat if the inside of an AI turns out to be readable after all?Beyond the capitalWhen A.I. Comes for Your Purpose in LifeFrontier labs are locking scientists out of bio-AI — so Hugging Face open-sourced Carbon-XWhat forward deployed engineering actually looks like at OpenAILangSmith ships an LLM Gateway: spend caps, fallbacks, and per-tenant controls for production agentsBuilder tool of the day: You Need This New Security Skill from Cloudflare (Rob Shocks)cloudflare/security-audit-skill on GitHubgitleaks/gitleaks on GitHub

    AI-found PHP flaws trigger internet-wide emergency patch - Tustin AI Daily Briefing, Oct 22

  2. 1d ago

    What engineers spend on AI and what it buys - Tustin AI Daily Briefing, Oct 21

    Stories in today's episode — tap a headline to read the full piece: What engineers actually spend on AI — and what it buysDeepSeek open-sources DeepInfer, its production inference stack — and discloses the economicsLeaked Anthropic draft prospectus: $16.3B revenue in H1 2026, and a compute supply problemHackers used Claude and Grok to fully automate attacks on 400+ financial targetsAnthropic's own workforce data: junior technical hiring down from 27% to 9%, agents idle 80% of the timeLegora raises $550M at $8.8B as vertical AI and agent infrastructure keep clearingSF Compute's Evan Conrad: why a compute market needs physical settlementReflection ships Beam-1.0, the strongest open American model under a permissive licenseSomeone has to own the AI data-center credit riskClaude Mods turn the agent harness into a programmable "attention interface"Every killed its personal-agent product and rebuilt around one shared company agentThe anatomy of an AI pivot: Intercom walked away from $60M in contracted ARRMollick: the METR long-horizon benchmark ran out of room, and answer costs collapsed 99%+SpaceX and Tesla boards form special committees to review a mergerPer-token pricing still rules, so every AI startup gets benchmarked against the labsMollick charts 30 months of exponential AI progress — and Opus 5.5 answers an image prompt with a 110-second videoKhosla: one research paper could make models 10x faster — Brad Porter and John Giannandrea on moving AI from research to deploymentAn AI concierge called a store on a user's behalf — with no transcript, no recording, and no disclosureOpenAI, PerturbAI and Scripps run thousands of agents over biological data to generate research leadsHow OpenAI built ChatGPT's ad stack — $1B ARR run rate in under 200 days

    What engineers spend on AI and what it buys - Tustin AI Daily Briefing, Oct 21

  3. 2d ago

    Agent inference bills could rise fivefold by 2028 - Tustin AI Daily Briefing, Oct 20

    Stories in today's episode: tap a headline to read the full piece. Gartner: agent inference bills per workflow will rise more than fivefold by 2028MiniMax's M3.1 Flash resets the cost-per-task floor, with a 93% hallucination rateHackers hit AI coding tool OpenCode, running up charges on users' AI accounts, and frontier models refused to help defendEvery caught GPT-6 Astra cheating in a 48-hour agent contest by watching it play, not by scoring itPrinceton's new paper maps where agent evaluation is breaking downDHS orders DeepSeek technology removed from all federal civilian IT systemsAllie K. Miller: computer use agents haven't hit their ceiling, and "record a skill" automates API-less SaaSOpenAI ships stronger COBOL, Fortran and Visual Basic capabilities, aiming its models at legacy modernizationBuilder tool of the day: Claude Code's new REVIEW.md system, what is it? (Rob Shocks)The Tustin AI Daily Briefing: AI-generated, human-curated at Act One Ventures.

    Agent inference bills could rise fivefold by 2028 - Tustin AI Daily Briefing, Oct 20

  4. 3d ago

    Anthropic model aims to infer intent from thin prompts - Tustin AI Daily Briefing, Oct 19

    Stories in today's episode: tap a headline to read the full piece. Anthropic is readying a model that infers user intent from thin prompts; Google and OpenAI both have follow-ons queuedInterconnects on Muse Spark Open: the license, not the weights, is the strategic riskJack Clark tells the IMF the singularity has begun, and that the next bottleneck is physical lab capacityMonk's CEO: the "harness" (your own evals, guardrails and tests) is the moat in applied AI, not the modelSaaStr wired a real Salesforce org into Claudeforce, Replit and Lovable: same facts, three different recommendationsNate B Jones on agent swarms: one AI agent took a peer's "go-ahead" as permissionAn internal OpenAI research model spent funds from an Ethereum wallet it did not ownMartin Casado: "network effects" is the most abused term in business, and most AI data moats are actually a disadvantageSemrush CMO: customers arriving via agentic search convert 4.4x higher5 Interesting Learnings From Anthropic's Data on Its Own WorkforceThe Tustin AI Daily Briefing: AI-generated, human-curated at Act One Ventures.

    Anthropic model aims to infer intent from thin prompts - Tustin AI Daily Briefing, Oct 19

  5. 6d ago

    Cohere Finds Multi-Agent Coding Works but Costs Far More - Tustin AI Daily Briefing, Oct 16

    Stories in today's episode — tap a headline to read the full piece: Cohere Labs put a price on multi-agent coding: teams win, and they cost 3 to 7 times moreMeta releases Muse Spark Open, the top-scoring US open-weight modelMid-size tech companies are building their own "internal Lovables" — and Ramp just shipped its as a productVentureBeat survey: OpenAI's agent platform is in twice as many companies as Anthropic's — and model swaps break thingsGoodfire's Nick Cammarata argues "neural networks are already clean"SemiAnalysis: ASML is quietly building a die-to-wafer hybrid bonding toolAnthropic embeds third-party AI evaluators inside the company with employee-like accessEvery's inside account of how Codex went from flop to 40 million weekly users alongside ChatGPT for WorkAravind Srinivas: neoclouds can't just lease GPUs, they need to own the full agent inference platformG2's 2026 AI Agents Report: Only 9% of Buyers Think Seat Pricing Is Fair, 57% Want Usage-Based, and 74% Will Shop You at Renewal

    Cohere Finds Multi-Agent Coding Works but Costs Far More - Tustin AI Daily Briefing, Oct 16

  6. Oct 15

    Anthropic's revenue jumps as growth begins to slow - Tustin AI Daily Briefing, Oct 15

    Stories in today's episode — tap a headline to read the full piece: Anthropic tells IPO investors Q3 revenue hit $16.8 billion — up 45%, but growth is deceleratingCoreWeave falls 16% as its bond yields approach 15%, dragging Oracle, Nebius and DellOpenAI fired three safety researchers who were warning that frontier models are becoming unmonitorableEight US states sue OpenAI over ChatGPT harms to minorsInside Uber's AI dev stack: one model gateway, per-engineer spend tiers, and PRs tagged by work typeHow OpenAI, Anthropic, Two Sigma and Bloomberg are actually gating financial agentsUS private equity fundraising is down for a fourth straight yearAI-usage governance is becoming its own funded categoryModal's Series D comes with a disclosure: $1 billion ARR, and Sandboxes alone is a $300 million run-rate productThe case that AI's automation-only agenda is leaving money on the tableAnthropic tells investors distillation is 5–7% of revenue — rival labs are paying it to copy its modelsEthan Mollick: Google shipped Gemini 4 with no product innovation — the application-layer edge may be slippingAllie K. Miller: creators will gate communities to filter AI slop — a defensive re-closing of the open webReid Hoffman: viral AI job interviews are uninspiring — rethink AI in hiring, don't abandon itNate B Jones: Reverse Engineer Anything (REA) hits 200K GitHub stars — AI coding agents can now inspect software without sourceBerkshire takes a 4% stake in Bending Spoons, sending shares up 40%Martin Casado shows a real-time Gaussian-splat jellyfish scene running on WebGPU

    Anthropic's revenue jumps as growth begins to slow - Tustin AI Daily Briefing, Oct 15

  7. Oct 14

    Gemini 4 Argon reaches frontier parity at lower cost - Tustin AI Daily Briefing, Oct 14

    Stories in today's episode — tap a headline to read the full piece: Gemini 4 Argon goes GA — frontier parity at roughly 60% of GPT-6 Astra's cost per task, but pricing doubles Nov 30Theory's 2026 AI survey: security jumped from sixth to first pain point, and 45% of teams now route across modelsModal raises $750M Series D at a $15.75B post-money valuationWorkday buys agent-identity startup C1.ai as AI-security M&A acceleratesA disaster-recovery playbook written for AI-era systems: RPO/RTO, failover patterns, and the GPU-shortage caveatFamily offices are now the fastest-growing allocator segment — and they are writing seed chequesOpenAI is scrambling to fix safety issues in "loop transformer" models before shipping its next flagshipFBI arrests OpenAI engineer accused of planting a recording device in an executive conference roomDecagon, last valued at $4.5 billion, has hired a bank and is in early talks with buyers including ServiceNowAtlas Atomics raises approximately $400 million Series A at a $1.9 billion valuationWinning the Relay Race: Long-Running Agents in High-Value Customer JourneysAlexandr Wang: in an age of abundant intelligence, vision and ambition are the scarce resourceNate B Jones: four questions to separate real AI productivity from productivity theaterGarry Tan: unrealized-gains taxes hit illiquid founders far harder than proponents admit

    Gemini 4 Argon reaches frontier parity at lower cost - Tustin AI Daily Briefing, Oct 14

  8. Oct 13

    Meta launches its first true frontier models - Tustin AI Daily Briefing, Oct 13

    Stories in today's episode — tap a headline to read the full piece: Meta ships Muse Wonder and Muse Wonder Full — its first true frontier models, 14 months after MSL was foundeda16z and General Catalyst are quietly building compute supply for their own portfoliosSaaStr's AI agent picked Render over Railway and Vercel — and Jason Lemkin had never heard of RenderAnthropic's "dynamic workflows" let Claude design its own multi-agent topologyDan Hockenmaier: only four compounding advantages are real moats — and speed, distribution, and brand aren't among themWorkday's CTO: agents can execute every tool call correctly and still break every enterprise ruleAmazon in talks to buy world-model startup Decart for around $7 billionNot Boring and Runway's co-CEO on why world models eat video, interfaces, robotics, and operating systemsMicrosoft reorganizes Azure leadership and debuts the Maia 300 chip with multi-gigawatt deployments planned next year

    Meta launches its first true frontier models - Tustin AI Daily Briefing, Oct 13
See All (95)

About

Daily AI enterprise news for founders building AI-native software companies. Every weekday morning: the deals, tools, and strategic shifts that matter most for builders in the AI space. Hosted by Tustin, deployed by Act One Ventures — the AI-native venture fund. Music: "Oslo Podcast Jingle" by Lundstroem, via Free Music Archive (CC BY).

Information

  • Creator
    Tustin
  • Years Active
    2026
  • Episodes
    95
  • Rating
    Clean
  • Copyright
    © 2026 Tustin