In partnership with
Happy Thursday 🫡 The theme this week: control and clarity in tech. Whether you’re shipping AI tools, building consumer apps, or just trying to outsmart reservation scalpers, this week’s stories will sharpen how you build, market, and monetize in 2025.
In this issue:
Genie 3 can build playable worlds from a single text prompt 🕹️
OpenAI just gave you GPT in your backpack—no API required 💻
The new startup moat: insight, not speed ⚡
How two founders beat NYC’s reservation scalpers to $200K ARR 🍽️
The 2025 AI marketing playbook that actually drives revenue 🎯
An Entire Month of Videos Before Lunch
Tired of the post-every-day grind?
Syllaby.io automates your entire content workflow. All you need is a topic—our AI does the rest.
✅ Get daily viral content ideas
✅ Auto-generate scripts tailored to your niche
✅ Instantly create faceless videos
✅ Bulk schedule across all your platforms
Syllaby is perfect for coaches, creators, and marketers who want to grow without showing their face or spending hours editing.
Genie 3: Google DeepMind’s leap into real-time world generation
TL;DR: Google DeepMind just unveiled Genie 3, a general-purpose world model that can generate fully interactive environments from a single text prompt. These 720p, 24fps worlds maintain consistency for about a minute and let users drop in new objects, characters, or rules on the fly. The model tracks past actions in real-time to ensure physics, visuals, and interactions feel coherent—making it a powerful tool for agent training, generative media, and open-ended simulation research.
🎮 Worlds from words – A single prompt can generate explorable 3D spaces with natural physics like water flow and dynamic lighting.
⏱️ Memory matters – Genie 3 remembers up to a minute of world history, simulating what happens next based on user actions or inserted objects.
🛠️ Interactive by design – Unlike its predecessors, users can change the environment in real-time—adding new characters, altering terrain, or shifting world dynamics.
🤖 Agent playground – DeepMind is testing Genie 3 as a training ground for its SIMA generalist agents, simulating goal-driven interactions in synthetic worlds.
⚠️ Not quite “The Matrix” – Current limitations include short session durations, limited agent action spaces, and imperfect multi-agent interactions.
Real-time, coherent world simulation unlocks two huge opportunities: (1) infinite training environments for embodied AI agents and (2) a creative sandbox for interactive media and design. If DeepMind can extend session length and agent capabilities, world models like Genie could become the backbone for next-gen AI systems—where agents learn safely in synthetic realities before stepping into the real one.
OpenAI drops open-weight GPTOS models: AI in your backpack
TL;DR: While we’re all waiting on GPT-5, OpenAI just released GPTOS 20B and 120B, the first cutting-edge American open-weight models you can download and run locally. The 20B model runs on most modern laptops, while 120B needs a beefier machine (~60 GB RAM). Running locally means no API costs, full privacy, and complete control—plus a peek into the model’s chain of thought. This is a major shift toward decentralized, user-owned AI.
💻 Local-first AI is here – GPTOS 20B runs on most laptops, while 120B is built for beefy workstations. By skipping the cloud, users get instant access, no latency, and total independence—a major shift from subscription-bound AI.
🔒 Privacy you can trust – Running locally means your data never leaves your device. Sensitive work, personal chats, or creative projects stay private, free from corporate logs or outside oversight.
🔍 Transparency plus control – Open chain-of-thought reasoning lets users see how the AI thinks. Pair that with full customization—from removing restrictions to fine-tuning outputs—and these models become personalized AI assistants.
💸 Free scale, community momentum – Local use kills API costs and enables always-on workflows at essentially zero marginal cost. This release also pushes U.S. AI toward open-weight models, sparking community-driven innovation.
Expect a surge in DIY AI apps, custom assistants, and locally deployed agents as developers seize this new autonomy. If the trend sticks, cloud-only AI could start to look like dial-up internet.
“Build fast and break things” might be dead
TL;DR: In 2025, shipping fast is no longer a competitive edge. AI and no-code tools let anyone clone multi-million-dollar products in a weekend—like Michael Luo, who replicated a $16B business model in just two days. The real advantage now lies in knowing what to build before the market moves. Kushank shows how founders can use AI-powered workflows to track competitors, mine customer complaints, and generate weekly market intelligence reports that guide product decisions based on genuine demand, not by following trends on X.
🚀 Product cloning is trivial – AI and no-code killed speed as a moat. Winning now comes from unique insights, not how fast you ship.
⚠️ Market need still kills startups – 90% of failures trace back to solving the wrong problem. Faster building doesn’t fix that.
🔍 Automated intel is the new edge – AI workflows pull competitor updates, user complaints, and market gaps into Slack, turning scattered signals into a living roadmap.
👥 Communities are goldmines – Reddit, forums, and niche groups reveal real frustrations. GEI used this approach to pivot to academic writers, driving 7,500% growth and $10M ARR—and you can replicate it today with Zapier, Perplexity, and LLMs.
In 2025, building fast is easy, but building right is rare. The startups that win aren’t the first to ship; they’re the first to uncover real, underserved problems. By turning community chatter and competitor moves into automated intelligence, founders can spot market gaps early, focus on niches that actually convert, and build products competitors can copy but customers won’t leave.
Your boss will think you’re a genius
Optimizing for growth? Go-to-Millions is Ari Murray’s ecommerce newsletter packed with proven tactics, creative that converts, and real operator insights—from product strategy to paid media. No mushy strategy. Just what’s working. Subscribe free for weekly ideas that drive revenue.
How two founders beat NYC’s reservation scalpers to $200K ARR
TL;DR: Table One is a membership-based app helping New Yorkers snag impossible restaurant reservations by beating scalpers at their own game. In under a year, the bootstrapped startup hit $200K ARR with 99% margins—without a single dollar spent on ads. A viral New Yorker article and organic word of mouth fueled rapid adoption, while the founders quietly built a lean, community-powered business.
🍽️ Consumer-first in a restaurant-first world – Existing platforms like OpenTable optimize for restaurants, not diners. Table One flips the model, giving members faster access to real openings.
🚫 Taking scalpers out of the equation – By monitoring reservation systems and pushing real-time alerts, the app undercuts the gray market where tables are resold at 10x the price.
📈 $200K ARR with 99% margins – The founders bootstrapped, spent almost nothing on ops, and focused on retention (89%) and conversion (40% from free to paid).
📰 Press as a growth accelerant – A single New Yorker feature catapulted downloads from 7K to 21K in a month, showing the power of timing and narrative in consumer apps.
🤝 Community as capital – Their funding came via a self-built portal where users could invest directly, aligning incentives and deepening loyalty.
Table One’s success comes from exploiting a market inefficiency with minimal friction. By layering on top of existing reservation platforms instead of replacing them, the founders avoided heavy technical lift and restaurant resistance, creating a business where both consumers and incumbents win. It’s a blueprint for how small, scrappy apps can build high-margin, subscription-first consumer businesses without burning cash on ads.
The AI marketing playbook every brand will need by 2025
TL;DR: Marketing is being redefined by AI. Not by replacing marketers, but by empowering them. The future belongs to teams that pair human creativity with AI-driven execution: think LLM-optimized search, micro-targeted campaigns, interactive experiences, and automated sales enablement. The hosts of Marketing Against the Grain lay out a six-part AI marketing strategy that turns old-school content into high-performing, AI-native campaigns.
🤖 Humans + AI = marketing edge – Marketers who partner with AI gain speed and creative leverage without losing the brand’s voice.
🔍 AI search + micro-audiences win – Optimizing for LLMs and running hyper-targeted campaigns beats old-school SEO and broad ads.
🎥 Video + interactive content convert – Short-form videos, calculators, and lightweight tools drive engagement and leads faster than static blogs.
🎯 Intent-driven automation scales sales – AI turns buyer signals into personalized outreach, boosting conversions without bloating headcount.
Throwing content at your audience won’t do it all in 2025. Marketers need to be making smarter, AI-augmented content that’s personalized, interactive, and discoverable in an LLM-first world. SEO is evolving, campaigns are shrinking, and sales and marketing are converging around intent-driven automation.
Want to sponsor this newsletter? This newsletter goes out to thousands of creators, builders, and early-stage founders every week. If you’d like to get your product or brand in front of them, you can book a sponsorship slot here 👉 Become a Sponsor
