How large ad budgets and AI token costs shape growth and design

At our summit on Monetization a few weeks ago, Daniel Chalef talked about how Zep AI (YC W24) uses limits. He said something that has stuck with me. That he views limits as "the connective tissue between PLG adoption and enterprise expansion." I love this framing. And at Schematic, we believe this type of thinking can help our customers make a ton of money. But you need the tooling to manage limits. Manage limits? Track & monitor limit usage. Webhooks and alerting, so you're notified that a customer is approaching or exceeding a limit threshold. Over time, as this becomes a core selling motion for your business, you need real visibility into the lifecycle of a customer's relationship to their entitled limits. And that's what we're launching shortly (screenshot teaser below). Take a customer. Look at their feature or limit usage with granular time stamps. Look at how their usage is evolving. Take action on helping them expand or rightsize their plan. Retain more customers. Expand more customers. Understand the behavior of your customers, as well as ideal behavior and how that maps to what they're paying you. Jon Brasted has been leading this sprint for us. Super excited to release this to our customers.

First look: VS1 Finance Demo VS1 brings AI and DeFi together on XRPL, combining strategy optimization, risk controls, and data-backed decision-making. It’s built for accessibility, capital efficiency, and continuous shipping, with a focus on real-world performance rather than hype. Our CTO, George Kiknadze, walks through the platform and its features, which are already available to test with an access code. Live now: swaps and limit orders on XRPL with fast execution, low slippage, and built-in risk tools designed for trading in volatile markets. The VS1 AI agent provides real-time analytics across XRPL activity, from transaction volumes and liquidity pools to token trends, and goes further by proposing personalized portfolios aligned to individual goals and risk. The experience is designed to move seamlessly from insight to action. Users can set preferences, receive tailored recommendations, and execute trades directly within VS1. A clean dashboard, intuitive navigation, and integrated portfolio management make it equally effective for beginners and professionals. The demo video is attached. Which aspect stands out most to you: the AI agent, XRPL swaps, or the overall UX?

"Unlocking Financial Efficiency Through Innovative Technology" In the fast-paced financial environment of 2025, three apps have revolutionized personal finance management: Cleo, Rocket Money, and PocketGuard. Cleo, facilitated by AI chatbot technology, tailors savings and budgeting to individual users. Rocket Money identifies and cancels unused or overpriced subscriptions, further optimizing expenses by negotiating bills. PocketGuard provides valuable spending insights and subscription management through its user-friendly zero-based budgeting features. By automating saving, providing personalized budgets, and highlighting unnecessary expenses, these innovative apps are enabling users to save hundreds of dollars annually. In this era of technology-driven solutions, are we harnessing these tools to their maximum potential? How is your personal finance management evolving in response to these advancements? I am curious to know your thoughts.

For the past year, the conversation around AI has been dominated by hype. But agents are only truly useful when they can be shipped to perform real work. elizaOS v2 is our answer: a production-ready framework for building and deploying agent-native workflows. Real builders are already using v2 to create tangible value across three pillars of utility: 𝗙𝗼𝗿 𝗗𝗲𝗙𝗶: Builders are deploying agents that automate complex financial strategies, manage protocol-owned liquidity, and rebalance portfolios on-chain. 𝗙𝗼𝗿 𝗢𝗿𝗴𝘀 & 𝗗𝗔𝗢𝘀: Organizations are streamlining their operations with agents that manage grants, coordinate proposals, and empower their communities with automated tools. 𝗙𝗼𝗿 𝗧𝗲𝗮𝗺𝘀 & 𝗖𝗿𝗲𝗮𝘁𝗼𝗿𝘀: Creative and technical teams are enhancing their productivity with agents that coordinate complex projects and manage shared knowledge bases. The proof is in what's being built. Our v2 Builder Spotlights start this week, where we'll go live with the teams shipping these workflows. First up is Senpi, a pioneer in agent-native DeFi.

Today's AI market is dominated by thin UI generators that connect with existing backend technology. Their only moat seems to be hype. I belief the real long-term game will be played in the backend. #angelinvesting #angelinvestors #investor #investors #investing #angelinvestor #investments #invest Picture: live admin portal by https://empowerd.dev CMS

What if your AI could autonomously settle a multi-million dollar international invoice in seconds, for pennies? This isn't science fiction; it's the convergence of Web3 and intelligent automation. We're seeing AI agents pioneering transactions on decentralized ledgers, completely reimagining B2B finance. This powerful synthesis is disrupting legacy systems by eliminating costly intermediaries and slashing settlement times from days to moments. It's about decentralizing financial control while optimizing global cash flow for unprecedented efficiency. How does your operational strategy shift when cross-border payments are no longer a bottleneck, but a real-time competitive advantage? #Web3Innovation #AIinFinance #CrossBorderPayments #EnterpriseDeFi #AgenticAI #FutureOfFinance #AutomationStrategy

With all the massive product launches by Big 5 Tech (AI Browsers, Image Generators, MCPs, etx), you can tell that LLM is not an answer to modern B2B or B2C problems - ecosystem is. They will keep building or buying vertical players to expand. The best you can do, as a startup or scaleup, is to build Narrow AI practical application software. Grind.

Distribution is the new differentiator. While the AI world is captivated by model capabilities, the data shows a more fundamental truth: go-to-market strategy is outpacing technology in driving user adoption. My takeaway from the latest traffic data: Browser-based agents are scaling roughly 10× faster than desktop agents, which in turn grow ~10× faster than agent orchestration platforms. Just look at the trajectory of Perplexity Comet. In the last quarter alone, it exploded from zero to over 2.5 million monthly visits. This isn't just product-market fit; it's channel-market fit. By embedding directly into the user's primary workflow—the browser—these agents eliminate the friction of installation and context switching that desktop and platform solutions face. It’s a powerful lesson for every founder and investor in the space: the most sophisticated agent technology may lose to a competitor with a frictionless, integrated distribution channel. Is distribution becoming the most defensible moat in the AI agent race?

How do you build a product so good that billion-dollar companies don't stand a chance? The other day, someone left a comment in our discord server telling us that not even the giants could compare. So how does this happen? It's called being AI-first. The reason startups are hot right now is AI has introduced an entirely new playing, allowing for tech to do things impossible only a year before. However, people are approaching AI incorrectly: Replit, Cursor, Lovable, bolt; they are all trying to take existing systems and tacking on AI features on top to make it a couple of percentage points better. When we built vly.ai (YC F24), we thought about AI from the ground up: how do we build a system that AI would perform best in? How do we architect something AI-native that would make it impossible for our competitors to even come near us? We worked extensively for over a year to build out the infrastructure, optimizations, tools, and features that made this a reality. Now, for any prompt, I can build it on vly, but never anyone else. Our own internal slack clone. Realtime ticket-tracking CRM. Multiplayer typeracing game for programmers. All in just one prompt. No one knows this yet, but the big players are about to be replaced by us companies going AI-first.

Every chain: “We’re decentralized.” Node distribution: 12 wallets, same building. We’ve all seen it. “Decentralization” as a marketing line, not a design principle. Infra isn’t truly decentralized until no one entity can pull the plug, rewrite history, or decide who gets to build. That means diversity in node operators, geography, and governance. Not just multisig screenshots and fancy infographics. At Go!, we’re obsessed with making decentralization real. Permissionless node participation, modular validation, and incentives that attract builders, not just speculators. Because if the next generation of AI agents, tokenized assets, and real-world markets are going to run on-chain, they can’t run on trust. They need proof. Comment your thoughts or DM me if you’re building real infra, not buzzwords.