// daily-digest · wed 06 may 2026 · morning refresh · last 72h focus
Anthropic ships the Wall Street stack. Klarna ships stablecoins on Tempo.
A day after the Pentagon split, Anthropic counter-launches Opus 4.7, ten financial-services agents, full Microsoft 365 add-ins, and a Moody's data deal — the bank-software market is being rewritten in real time. Plus Klarna issuing a USD stablecoin on Stripe and Paradigm's Tempo chain, Sierra at $15B as enterprise-agent gravity consolidates, Stratechery on Amazon's quiet logistics coup, Meta scanning bone structure to flag minors, Sean Goedecke's plain-English Tempo FAQ, Jensen Huang on the job-creation pitch, Denver home prices that won't quit, image models eating chatbot growth, and a wildcard from NEJM that may have just retired your knee surgery.
— refreshed for you, in 10 items.
idx
title
type
[01]
article · anthropic · may 05
Fortune — Anthropic deepens push into Wall Street with new AI agents, full Microsoft 365 integration, Moody's data partnership
Why for you: Ten ready-to-run agent templates — pitchbook builder, earnings reviewer, financial model builder, GL reconciler, month-end closer, statement auditor, KYC screener, more — riding on Claude Opus 4.7 (now leading Vals AI's Finance Agent benchmark at 64.4%). New connectors to FactSet, S&P Capital IQ, MSCI, PitchBook, Moody's, IBISWorld, D&B, Verisk. Claude is now a first-class citizen inside Excel, PowerPoint, and Word. The "consulting venture" from yesterday is not vague — these are the products it's selling.
Anthropic: Introducing Claude Opus 4.7 (model card + benchmarks)
[02]
article · stablecoins · may 04
The Block — Klarna announces a USD stablecoin on Stripe and Paradigm's Tempo blockchain
Why for you: Klarna isn't a small player and Tempo isn't a side project. The pitch: cut interchange and FX out of cross-border BNPL, settle on a chain whose dedicated payment lanes target one-tenth of a cent per transaction. The agentic-commerce story you've been tracking now has a real merchant-side flywheel: rails (Tempo), tokens (USDK), and an issuer (Klarna) all on one stack. The interesting question is whether Visa/MC respond before Q3.
[03]
article · stratechery · may 05
Stratechery — Google Earnings, Meta Earnings (and Amazon's quiet logistics coup)
Why for you: Thompson buries the lede: the most consequential news of the week wasn't AI — it was Amazon Supply Chain Services opening Amazon's "full portfolio" of fulfillment, freight, and last-mile to non-Amazon sellers. FedEx and UPS got tagged on the announcement. Meta's NDRC fight over the Manus reincorporation gets re-read as Beijing reactive, not strategic. Standard Stratechery payoff: the AI-narrative crowd was looking the wrong way.
[04]
article · enterprise ai · may 04
TechCrunch — Sierra raises $950M as the race to own enterprise AI gets serious
Why for you: Bret Taylor's Sierra closes a Tiger Global / GV-led round at $15B post. The same week Anthropic + Blackstone + Goldman announced their $1.5B forward-deployed venture, and OpenAI sealed The Deployment Company at $10B. Three different bets on "enterprise AI = vertical agents + service layer." Worth tracking who wins which verticals — Sierra has the customer-support beachhead, but financial services is now actively contested [01].
[05]
article · ai policy · may 05
TechCrunch — Meta will use AI to analyze height and bone structure to identify if users are underage
Why for you: Photos and videos get scanned for visual cues that someone is under 13; if flagged, account deactivated unless the user clears Meta's age-verification flow. The deeper trend: AI is moving from "produce content" to "audit users," and biometric-by-photo is suddenly normalized. Note for your kid-and-internet conversations — and for thinking about what "consent" means when the inference happens on existing photos already on the platform.
[06]
article · sean goedecke · recent
Sean Goedecke — An unofficial FAQ for Stripe's new "Tempo" blockchain
Why for you: Cleanest plain-English explainer of why Tempo exists, why fees are paid in stablecoins (no native token), why Stripe gets to skim fees on flow it routes anyway, and why "blockchain for payments" is finally a real phrase rather than a marketing one. Pair it with [02] — together they're a 20-minute primer on the layer that's about to handle agentic commerce settlement.
[07]
article · ai labor · may 04
TechCrunch — Jensen Huang says AI is "creating an enormous number of jobs"
Why for you: Standard Jensen optimism, but the framing is now load-bearing for Nvidia's policy position with Washington. Worth holding next to Sean Goedecke's "software engineering may no longer be a lifetime career," Cherny's "coding is solved," and Friday's DeepMind union vote — three other 2026 takes on the same question. The vendor with the most to gain from "AI is additive" is, of course, the one selling the picks and shovels.
[08]
article · ai product · may 04
TechCrunch — Image AI models now drive app growth, beating chatbot upgrades
Why for you: Appfigures: image-model launches generate ~6.5x the downloads of language-model upgrades. Nano Banana drove +22M Gemini installs in 28 days; GPT-4o image drove +12M for ChatGPT. Useful counterweight to "the model is the product" — for consumers, the model that ships a viral image style still beats a benchmark jump. If you're sizing distribution, this is the new lever.
[09]
article · denver · may 05
Denver Gazette — Denver home prices open up gains despite high interest rates
Why for you: Metro median resale at $605K in April, second straight month up despite rates and an inventory glut everyone said would crack the market. Insurance is now the line that's actually moving — average CO homeowners premium ~$4,100/year, up 137% in a decade, reshaping who can afford which neighborhood. Useful baseline for any buy/sell/refi conversation in the next quarter.
[10]
[wildcard] · medical reversal · may 05
ScienceDaily — Common knee surgery found ineffective, may make things worse (10-year FIDELITY follow-up)
Why for you (off your normal lanes): NEJM 2026: 10-year follow-up of the Finnish FIDELITY trial, where degenerative meniscus tears were randomized to arthroscopic partial meniscectomy vs. sham surgery. The surgery group fared worse — more symptoms, faster osteoarthritis progression, more reoperations. Yet partial meniscectomy is one of the most common orthopedic procedures in the US. A clean example of "medical reversal" — practice that survived because the placebo arm was uncomfortable to run. Useful both as a runner and as a reminder that the standard-of-care prior is not the same as the evidence prior.