πŸ’ͺ Mighty Brief

Friday, March 20, 2026 Β· 5:00 AM CT
⚑ TL;DR β€” Top Stories
Greg Brockman back to lead the consolidation. Fidji Simo: "phase of refocus." Targets engineering/business customers. Karri Saarinen (Linear): keep Codex separate. β€” WSJ
Not open weights. Trend: frontier Chinese models closing the gap but going proprietary. β€” VentureBeat / Techmeme
8% single-day drop. Market pricing vibe design as existential threat to traditional design SaaS. Canva quietly waiting for the right window. β€” The Information
Blue Origin files for 52K orbital satellites for "Project Sunrise" β€” AI data centers in space
Seeking FCC approval. The compute-in-orbit thesis gets serious. β€” WSJ via Techmeme
Walmart embedding Sparky directly into ChatGPT + Gemini instead. First major data point on agentic commerce underperforming. β€” Wired
Top desire: professional excellence (18.8%), personal transformation (13.7%). Healthcare workers: "the pressure of documentation has been lifted." β€” Anthropic
Alibaba + Tencent lost $66B combined as AI monetization plans stay unclear
Bloomberg. The gap between AI capability and revenue model is still very real. β€” Techmeme
SEC approves Nasdaq rule for tokenized securities trading β€” pilot program launching
The Block. Institutional crypto infrastructure moving forward regardless of crypto volatility narrative. β€” Techmeme
πŸ—οΈ Blank Metal
πŸ“Š Pipeline
DealStageStatusValue
PowerPusherSOWWARM Ryan Blesi following up β€” Dan promised SOW "tomorrow" on 3/18. Overdue.$78-135K
TufinActive DeliveryHOT Working sessions #1 + #2 done yesterday. WS #3 today 1:30-3:30 PM (SF data connection + use case prioritization)Active
Tungsten AutomationLead→DiscoveryWARM Elli sent next-steps email to Ryan Hubbard. Materials ETA: early next week. Scheduling mid-to-late next week.TBD
NeoGovDiscoveryHOT $150K–$1.5M. Discovery call TBD.$150K+
PaceLoanGroupProposalHOT ~$100K, 3 weeks no reply. Nudge needed.~$100K
TalkiatryProposalHOT $400M telepsych company. $400K+ opportunity.$400K+
BBGDiscoveryHOT (stall) Samples 19 days late. $391K deal.$391K
Zivian HealthSOWWARM Partner POV needs sharing.TBD
GreatHillPartnersSOWWARM Ingest: SOW email active yesterday.TBD
πŸ‘₯ Team Activity (last 24h)
Git commits: Granola auto-sync (2), PowerPusher email ingest (Ryan Blesi thread + attachments), Tungsten Automation next-steps email from Elli
Elli β†’ working Tungsten next steps, coordinating Ocean Sound asset reuse for Tungsten deck
⚑ Dan's Plate
πŸ”΄ PowerPusher SOW β€” Ryan Blesi is waiting. Dan promised it 3/19. Now 3/20. Send or set expectation TODAY.
⚠️ OpenAI API auto-recharge failed (3/19) β€” check payment method / billing for dan.wick@blankmetal.ai
πŸ“… 8:30–9:00 AM β€” Busy block
πŸ“… 9:00 AM–12:00 PM β€” Services Work Session (Hold) β€” Mark, MJ, Elli, EJ
πŸ“… 12:00–1:15 PM β€” Busy block
πŸ“… 1:30–3:30 PM β€” Tufin WS #3: SF Data Connection + Use Case Prioritization (Zoom w/ Jordan Bothwick, Sheree Wong) β€” review use case spreadsheet beforehand
🧠 Munger Observer (Mar 19)

System Fragility β†’ Second-Order Thinking: A single LCM plugin bug triggered a 12-hour outage cascading across 7+ cron jobs. First-order fix: remove LCM. Second-order question: what other single points of failure exist in the stack? Resilience isn't "don't have bugs" β€” it's "no one failure should take down everything."

Commitment Overrun β†’ Inversion: Dan promised Ryan Blesi the PowerPusher SOW "tomorrow" after a Mar 12 revised alignment doc sat for a week. Invert: what happens if the SOW doesn't arrive? Ryan's already followed up once. The pattern β€” delay β†’ apologize β†’ promise tomorrow β€” is a trust-eroding loop. The fix isn't speed; it's never committing to a date without capacity to meet it.

Opportunity Cost of Meta-Work: Pipeline is hot (8+ active leads). Time spent cataloguing β‰  time spent closing. Awareness β‰  action.

"Invert, always invert." β€” Carl Jacobi (Munger's favorite)
πŸ”— Capture Themes (last 48h)
Agentic Payments Infrastructure
@jeff_weinstein (MPP/Stripe), @levie (zero friction micropayments), synthesis capture (token governance gap), WorldNetwork (AgentKit sybil resistance)
β†’ The plumbing for machine-to-machine commerce is being laid now. _M angle: enterprises need cost governance for agentic spend the way they needed Cloudability in 2015. This is a real positioning play.
Distribution as the AI-Era Moat
@lennysan (distribution is the new moat), @yoheinakajima (startup loop: generate→build→distribute→measure), @trq212 (Anthropic's internal skill marketplace = organic distribution)
β†’ In a world where AI equalizes coding output, how you reach users matters more than engineering effort. Wick.tech thesis confirmed from multiple angles.
The Middle Getting Priced Out
@j0hnwang (Figma -80% on Stitch), Walmart/OpenAI agentic commerce underperformance, Adobe Firefly custom models (public beta)
β†’ Traditional design SaaS, mid-tier commerce integrations, and legacy agency models are all being disrupted simultaneously. T5 in real-time.
Enterprise AI Governance Demand
@levie (20+ IT leaders: agents, governance, identity, token budgets), Anthropic 81K study (professional excellence #1 desire), Walmart/OpenAI restructure
β†’ _M ICP validated: regulated enterprises want agents + governance layer. Every conversation should surface token budget visibility as a core capability.
πŸ“§ Newsletter Inbox
🐦 X Signal
@paulg
Someone asked what past decade seems most similar to the current one. I said the 1930s. Great progress in technology, combined with corrupt, autocratic, populist political leaders.
β†’ x.com
@emollick (Ethan Mollick)
"There is some danger for the Big Three labs that they have run out of imagination and are now refining Codex/Claude Code/Antigravity... These were good UX for AI's use & limits today, but not great UX for the future of AI." Also: Google doing the weirdest UX experiments (Stitch, Flow, NotebookLM, AISOMA) β€” may pay off.
β†’ x.com
@pmarca
"OpenClaw and Pi together are in the top 10 of all time software breakthroughs." β€” Reacting to @chrysb calling OpenClaw "like the early internet."
β†’ x.com
@garrytan
"Markdown+LLM is code and for creating high value complex contextual and intelligence-required tasks it is much much better than any deterministic language"
β†’ x.com
@karpathy
Saw Project Hail Mary. Book is one of his favorites for alien portrayal depth. Movie stays faithful, adds superhero pacing. Rocky + bromance = well executed.
β†’ x.com
@clairevo
"I'm not gonna bet against Claude Code, but there is a very important lesson in product competition going on right now: The features β‰  the Product"
β†’ x.com
@steipete
RT: "If your $500K engineer isn't burning at least $250K in tokens, something is wrong."
@minnestar (Twin Cities)
Minnebar 20 session submissions now open! May 2 at the State Theater. Deadline: April 19. β€” minnestar.org
@pdouglasweather (Twin Cities)
MSP snow down to 2". Melting machine kicked in β€” should be mostly gone in the next few days.
🧩 Trending Skills (skills.sh)
Top movers relevant to your stack:
vercel-labs/find-skills (#1, 16K installs)
google-labs/stitch-design (↑ new)
google-labs/design-md (↑ hot)
anthropics/canvas-design
anthropics/mcp-builder
obra/dispatching-parallel-agents
obra/using-git-worktrees
vercel-labs/vercel-cli-with-tokens
mattpocock/write-a-prd
coreyhaines31/content-strategy
Notable: Google Stitch skills (design-md, stitch-design, shadcn-ui, react:components) are all new and moving fast β€” correlates with Figma/Stitch news.

Mighty Brief Β· March 20, 2026 Β· Generated 5:00 AM CT