Key Takeaways
- Adding MCP support to an existing API can be a single prompt, covering everything from authentication to plugin store listing.
- Seat-based SaaS pricing is becoming obsolete; credit-based metering for agent usage is the emerging model.
- AI agents will intensify competition in systems of record like CMS/CRM, not replace them.
- levelsio launched Hotelist, an AI-powered hotel rating aggregator, to counter inflated ratings from review deletion.
- Tibo's $12K sponsorship bet delivered 53 incremental trials, with only ~6 retained users needed to break even.
1. AI-Native SaaS: MCP, Pricing, and the Age of Agents
- Adding MCP support to an existing API can be as simple as a single well-crafted prompt. Arvid Kahl shared a detailed prompt covering authentication, documentation, analysis, testing, listing on OpenAI/Anthropic plugin stores, and a landing page, making MCP adoption dramatically easier for developers. — via 1
- SaaS pricing books need rewriting: seat-based pricing is obsolete because seats are now agents. Arvid Kahl argues for credit-based metering that charges for actual agent usage, a model better suited to AI-native applications. — via 1
- AI agents won't replace core systems of record like CMS/CRM. Jon Yongfook expects SaaS to consolidate around these tools, making those markets more competitive and more vertical as agent capabilities grow. — via 1
2. Hotelist Takes on Inflated Hotel Ratings
- Hotel ratings are widely distorted: platforms like Airbnb, Google Maps, and Booking delete negative reviews, driving averages up to 4.7, and paid fake-review services exist; Airbnb even removes user criticisms while keeping landlord attacks. In response, levelsio built Hotelist, which aggregates and normalizes ratings across platforms and uses AI plus real experiences to evaluate hotels—free and not commission-driven. — via 1 2 3 4 5
3. Growth Marketing: A $12K Sponsorship Bet and Organic Traffic Rankings
- Tibo spent $12,000 to win a bidding top spot, drawing imitators and accusations of flexing. The results, however, were strong: 44 trials on Friday, 38 on Saturday, and 31 on Sunday, versus a baseline of about 20 per day, or roughly 53 incremental trials. — via 1
- With a trial-to-conversion rate around 50% and an LTV near $2,000, Tibo needs only 11% of those 53 (about 6 people) to retain for the spend to break even. This provides a generous margin of error for paid acquisition decisions. — via 1
- Tibo also announced an organic traffic leaderboard: users connect their site, clicks are verified, and the internet decides the ranking. This adds a new transparency layer to site traffic measurement. — via 1
4. Practical Advice for Makers: Naming and Goal Setting
- Tw93 shared naming principles for products: names should be memorable, easy to write and spell, and roll off the tongue, while connecting to the product's meaning and giving the founder pride. He illustrates with examples like 妙言 (wonderful words), Pake (Creole for pack), and the Mole/Kaku/Waza/Kami series, noting that hard-to-spell names can hinder mass adoption. — via 1
- Nathan Barry defines the three parts of a useful goal: a date, a clear outcome, and a number. Vague goals, he argues, are useless for driving effective action. — via 1
