Alice ⧉ alisa.bnb

A great event @NewFriendTech on the sidelines of @EthereumDenver, organized by @EricSpivak. Thanks for having us, and a big thank you to @dcbk2la for leading our discussions. 💙 Great panelists from @Somnia_Network, @spaceandtime, and @nillion!

ANNOUNCING: PREDICTION MARKETS ON ARKHAM

We’re introducing a new GitHub Certified: Agentic AI Developer (GH-600). As AI agents become part of modern development workflows, this role-based certification focuses on how developers and teams operate, supervise, and integrate agents across the SDLC. If you’re already working with tools like GitHub Copilot or exploring agent-driven workflows, we’d love your input. Learn more and get involved. msft.it/6013vRHHZ

@Multihopper @solana @colosseum @Visa @JRA_xyz @kujcrypto Congrats to the team! This is big🔥

Well done to the @Multihopper cofounders for winning 3rd place at the @Solana @Colosseum Frontier @Visa track! @JRA_xyz and @kujcrypto pitched, with 50 companies competing. This was our first time we went out there pitching and came back with a win. A special thank you to @SuperteamDE for mentoring and hosting us through the hackathon. We were incredibly supported and wildly humbled and grateful. MultiHopper is programmable routing infrastructure for digital assets — think "SWIFT 2.0", rebuilt natively on-chain. Swift transfers $42TN of the world's FIAT every day between banks. We're building that infrastructure for crypto. MultiHopper lets you route assets across any wallet or contract on-chain with full control over path and timing — no custody, no signatures from hop wallets, no pooled funds. No mixing, tumbling to ridiculous liquidity requirements. Regulatory ready. Congratulations to all the other teams competing!

@0xMovez fair, jane street's moat is the latency and the data pipeline. MIT teaches the model, but the $750k pays for the infra around it.

Jane Street pays $750k/ year for quants who can answer how to use Stochastic Process and Markov Chains in quant trading. This 1-hour MIT lecture on probability gives you the same insights quants get paid $60K/month for. Bookmark & watch today. Then read the article below.

@DoctorYev mostly, the wrapper layer was always going to consolidate. What dies is anyone without proprietary data or distribution.

did Higgsfield just kill every AI Video startup? what's going on here

@usenaive well, if agents can incorporate companies, the next question is who underwrites them. KYC, banking, compliance, all wide open.

Naïve company formation just killed Stripe Atlas. Agents can now incorporate real companies, straight from the CLI. Get started by just prompting your agent: "Follow Naïve AI's docs to register and incorporate my company [name]" Your agent will then: > handle identity verification and KYC > form an LLC/C-Corp in any US state > file state documents and pull Articles of Incorporation & EIN letters Free CLI, free API — you only pay the no-markup $249 formation fee when your agent actually files (covers state fees, registered agent, EIN, all of it.) The "download a company" stack is real now. Try it ↓ 🔗

@katexbt right, every ad platform connecting to claude means agencies just lost their last moat. The $15k/month media buyer becomes a $30/month subscription.

🚨Cool tip for anyone in advertising: Meta officially connected Meta Ads to Claude The connector went live on April 29, 2026 Here is the site: mcp. facebook. com/ ads Setup takes about 60 seconds You go to claude settings, add it as a custom connector, authorize via facebook OAuth, and you're in Once connected, claude has full read and write access to your ad account (dangerous, but also highly cool) You can tell it what you're selling and who you're targeting, and it builds the entire campaign structure for you, especially if you feed it previous examples and statistics. Ad sets, targeting, copy, everything It can also monitor your ad pixel health, upload your product catalog, and generate performance reports 29 tools total, all free during beta. Think it can be done via Pro but maybe needs Max. Still WAY cheaper than the 3-5k usd range others would quote you before for this. It’s now a two-minute setup inside Claude

@igus_ai yeah, retail with an LLM stack just beat funds with bloomberg terminals. The alpha was always information asymmetry, agents collapsed the cost of synthesis.

Un estudiante chino viviendo en Japón convirtió $0,90 en $408.292 tradeando en Polymarket. Y casi nadie está hablando de ello. Su cuenta se llama “Gravia”. Solo llevaba 2 días en la plataforma. Pero lo más absurdo es esto: Dice haber construido un bot con Claude para tradear el mercado BTC UP/DOWN 5MIN. El sistema: • Obtiene datos en tiempo real desde Binance WebSocket + velas 5M • Cruza señales de TradingView + flujos de exchanges de CryptoQuant • Usa un force-graph con 100 nodos y 180 conexiones para detectar convergencia BULL/BEAR • Detecta retrasos entre el spot price y el CLOB de Polymarket • Ejecuta operaciones en menos de 100ms antes del repricing • Puede lanzar más de 1000 órdenes por segundo • Captura entre 0,3% y 0,8% por trade Pero aquí viene lo importante: El edge no está en “predecir Bitcoin”. Está en explotar microdesfases entre: • precio spot • señales del mercado • y repricing del order book de Polymarket Y según él, el bot directamente evita operar si: • no hay edge • la liquidez es baja • las señales se contradicen • o se alcanza el límite diario de riesgo También tiene controles bastante agresivos: • Riesgo por operación: 0,5% • Límite diario: 2% • Hard stop: -0,4% • Corre localmente • No usa GPU • No depende de cloud

@akshay_pachaar agree, self-evolving skills and tiered memory is the unlock. most AI agent startups are shipping wrappers when the real moat is recursive context.

@Must_be_Ash yeah, agent wallets are step one. Identity is the real bottleneck under all of this.

8 startup ideas up for grabs: 1. Agent Wallet with Programmable Spend Policies
Enterprise wallet for Perplexity Computer and Manus… This would be more like server wallet. The user never interacts with them or goes through onramp. The bill goes to their existing linked credit card. issues virtual cards and wallet with dynamic rules: spend caps per category, per-agent, per-time window, auto-approval thresholds, and audit trails.

Why now: Stage 2 is exactly where the approval gate is widening; no one owns the “agent spends safely” UX yet. 2. Agency w/ Billing Dashboard
help businesses transition. Include a dashboard to manage funds and off-ramp. 
Target: service providers. 
Why now: Stage 3 is exploding; AWS AgentCore already ships x402. Someone needs to make it dead-simple for the rest of the ecosystem. 3. Specialist Agent Marketplace
A directory workers, bundles and agents (video editor, research analyst, procurement bot, etc.). 
Why now: Stage 4 is literally “agents hiring agents”. This becomes the “Upwork for AI agents.” 4. Enterprise Agent Procurement OS
SaaS that lets company agents autonomously request quotes, negotiate with supplier agents, issue POs, and track fulfillment. 
Target: Mid-market and procurement teams. Starts with office supplies. 

Why now: Stage 4 B2B version hasn’t happened yet. 5. Headless Commerce Checkout Kit for Merchants
Drop-in SDK or marketplace plugins that turns any Shopify/eCommerce store into a fully agent-friendly endpoint 
Why now: Stage 2 already makes shopping headless; merchants are losing visibility unless they optimize for agents. 6. SEMrush or Ahrefs for the age of agentic commerce Show the gaps in the market and where the demands are. Which endpoints are getting action and exposing supply and demand signals 7. Paid Claude Code plug-ins. Skills that bundle x402 endpoints and don’t ask for API keys and boost Claude code’s abilities “turn your Claude plugin into a business that earns while you sleep.” Like micro SaaS but for Claude 8. Multi-Rail Agent Settlement Router. 
Smart router that gets rid of conversations and bridging The agent pays with whatever it has, and the merchant receives whatever it accepts no matter the rails

@victoria_olsina 💃💃💃

follow this legend!

@idohodl I think the tricky part is that usage-based feels fair only when the user understands the unit. With AI, the cost driver is often invisible, so pricing has to translate compute into something closer to value delivered.

Once you get used to the taste of usage-based pricing with AI tools, it's hard to go back to SaaS and pay monthly - That one point will kill SaaS revenues by a lot of margin - Or the cranking of value out of the product. It would even pump the revenues a lot just like how Claude pushed Anthropic revenues!!

@narek_gevorgyan @CoinStats Curious how other founders/operators think about this: when should AI products move from simple subscriptions to credits or usage-based pricing?

Most AI companies still haven’t figured out pricing. I asked @narek_gevorgyan, CEO of @CoinStats, how they think about monetizing CoinStats AI: subscription, credits, add-ons, or hybrid. His answer: AI pricing is still an open question. Usage-based pricing makes sense until one prompt turns into hours of compute. Video below 👇

@stackscans exactly. seats made sense when human access was the proxy for value. with agents, the buyer, user, cost driver and value unit can all split apart.

The SaaS pricing model we grew up with is dying. Per-seat pricing made sense when humans did the work. But AI agents let 1 employee do the work of 10. So why pay by seat anymore? What’s replacing it: → Usage-based pricing → Outcome-based contracts → AI agent consumption → Agentic Enterprise License Agreements (AELAs) In the AI era, customers won’t pay for software access alone. They’ll pay for results. If you’re still selling SaaS purely by seat in 2026, you may be optimizing for a market that no longer exists. How is your company adapting?

@MGIResearch the metering piece is underrated. most teams talk about AI pricing like it’s packaging, but the real bottleneck is knowing what was done, for whom, at what cost, and what business value it created.

Most companies are underestimating AI monetization. On May 21, 8AM PT, MGI Research explains why fewer than 5% of companies are ready for usage-based pricing and how AI is exposing weaknesses across billing, metering, and RevRec. Register: us02web.zoom.us/webinar/regist…

@Ekonixlab yep, bundles are the transition layer. AI features get stuffed into seats until compute cost forces a cleaner model: usage for cost, outcomes for value, commitments for predictability.

What is the real cost of SaaS shifting from seat-based pricing to usage-based pricing? I think it is already happening through bundles. Each “seat” gets more expensive as more AI features are added. The bigger risk is margin pressure as AI model providers like Anthropic and OpenAI eat more of the economics. That said, SaaS still owns the customer relationship. The models may power the workflows, but the apps closest to the user likely capture the most value.

@rahulrana_95 the seat model starts breaking once the “user” is an automated workflow, not a person. imo the harder shift is not lower headcount, it’s deciding what the unit of value actually is.

Wild tech paradigm shift: GitLab is restructuring, cutting management layers, and slashing its global footprint by 30%. Why? To "rewire" internal processes with AI agents for automated code reviews, approvals, and handoffs. The era of pricing software by the human "seat" is dying.

@adriankxyz this is exactly why pricing can’t live in a final_v7 deck. if the workflow is unwritten, the value unit is unwritten too, so agents automate chaos instead of creating pricing power.

Most companies are the opposite. The CRM says one thing. Slack says another. Pricing lives in "Final_v7_NEW." Best client context is in a few employees' heads. The real workflow is mostly unwritten. Then leaders wonder why AI agents don't transform the business.

I've probably eaten at @_Hillstone more than any other restaurant over the last decade. Somehow it always feels exactly the same in the best possible way. The lighting. The booths. The pacing. The frozen martini glass swap halfway through

Most AI/Web3 teams price too late. Pricing is not the last slide in your GTM deck. It should shape what you build, which user you prioritize, and what you refuse to give away. Watch more > (from @ETHCC)