ciao.fren

If you code with Cursor, Copilot, or any AI agent, this is for you. We built a plugin that pays you every time your agent is "thinking...." install once, keep working exactly like you do now. First 50 devs get early access. Reply with 🫰 and I'll dm you.

calling all AI agent builders if your agent spends time thinking, searching, analyzing, or running tools… we’re building something around that moment. looking for early AI products with: > a free tier > real users > meaningful inference costs > a few seconds of “thinking” time if this sounds like you, DM me. would love to talk.

Small wins 🥉 We wrapped our Product Hunt launch at #3 launch of the day. An exhilarating 24 hours; I'll write about the whole thing soon. But the real win wasn't the badge, it was that 8 brands were already working with us before we launched, and 36 more signed up and tried the product in 24 hours. What we're building is unique, interesting, and, most importantly, very much needed. Now back to the actual job, making brands visible in the age of AI :)

Our thesis at @scribble_dao is that AI engines cite real people talking about a brand more than the brand's own pages. Last week we got a clean look at it, and its wild how differently each engine behaves. The setup: a client (a DeFi protocol) was flat at zero visibility across the AI engines for over a week. We fixed the onsite issues first, helped build out the onsite content, then launched a creator bounty, briefed on the exact questions people ask about yield, liquidity and inflation. Creators published on all AI crawled platforms. Within 3 days, citations across ChatGPT, Gemini, Perplexity, Copilot and Grok. Then we tracked all five llms on the same question and checked where each pulled its sources from. Same question. Five completely different methods of answering. ChatGPT: leaned almost entirely on the brand's own site. Roughly ten owned-page citations, a couple of creator posts mixed in. This was the one engine where our onsite work did the heavy lifting. Grok: built the entire answer from a single creator post on Medium. One source. That's it. Perplexity: went wide. One creator post, then a spread of authority sites, sec gov, CoinGecko, YouTube, LinkedIn, a fintech publication. Classic search-engine sourcing. Gemini: almost wall to wall creator content. Counting everything, creators accounted for roughly three quarters of the sources. The owned site barely appeared. Copilot: same shape as Gemini. Creator posts on Paragraph and Medium up top, the brand's own page mentioned once. So on 4 of the 5, creator content carried the answer. On 1, the owned site won. Obvious caveats, because someone will ask: early numbers, small sample, one brand, one niche. But yes, the onsite work matters; it's literally what won ChatGPT. But the takeaway I keep coming back to: there's no such thing as optimising for "AI”. You’ve got to be everywhere. Each engine reads the web differently, so you need owned pages AND real people writing about you, because you don't get to pick which engine your customer uses. (We've been saying the creator part for months now. Nice to finally have the screenshots.)