Your customercustomer asks AI firstBe the answer.

AI assistants like ChatGPT, Gemini, and Perplexity don't pull from your website the way Google does. They pull from what's being said about you across the web: third-party articles, community posts, comparison pieces, and structured content that signals authority. Scribble runs the full loop. We audit which queries your brand is invisible on, generate citation-ready content on your own domain, then post bounties to 50,000 real creators who publish across every surface AI reads from. The result: AI starts seeing your brand in enough credible places that it names you in the answer. RocketX went from 2% citation share to 20% across 25 priority queries in three months.

Most brands see citation movement within 25 days of their first campaign going live. The first week is publishing. The second week is indexing. By week three, the AI models start picking up the signal. The compounding happens fast, once you're in one answer, the model is more likely to pull you into related ones. Three months in, brands running consistent campaigns see 5-10x citation growth on their priority queries. One campaign is a start. The real moat is built over quarters, not weeks.

We track and optimise for five: ChatGPT, Gemini, Perplexity, Copilot, and Grok (Claude coming soon). Each model weights sources differently. Perplexity pulls heavily from recent indexed content, ChatGPT favours authoritative third-party mentions, Gemini rewards structured on-domain content. Scribble's Audit dashboard shows your citation share per model so you can see exactly where you're strong, where you're losing, and which models your competitors are dominating. Most brands start with gaps on 3 of the 5. The goal is to close all of them.

Your top queries, the ones your competitors are winning, become the bounty brief. We send it to creators in our network who are scored by citation probability, not follower count. They write content structured around those queries and publish across X, Medium, Substack, Paragraph, Reddit, Youtube, Instagram and Quora. We pay when their piece gets cited by AI, not when it gets published. That one difference changes the quality of everything that comes in. 300+ creators compete per bounty. The ones who've landed in AI answers before go to the front of the queue.

X, Medium, Substack, Paragraph, Reddit, Youtube, Instagram and Quora, the eight surfaces AI models actively pull from when building answers. We picked these because they're crawled frequently, carry domain authority, and are treated as third-party signals by every major model. Reddit threads show up verbatim in Perplexity answers. Substack posts get cited in ChatGPT responses. Medium articles are indexed fast and treated as editorial. The combination means your brand gets signal across every platform a model might reference when someone asks about your category.

Your citation score is your share of AI answers for the queries that matter to your business. If 25 buyers ask ChatGPT "best cross-chain aggregator" and your brand appears in 5 of those answers, your citation score is 20% for that query on that model. Improving it comes down to three things: owned content on your domain that's structured for AI readability, third-party creator content pointing back to it, and consistency across platforms so the model sees your brand as the signal, not the noise. Scribble's dashboard tracks your score per query, per model, and shows you exactly which campaigns moved it.

Yes, and it compounds. Every piece a creator publishes on Medium, Substack, or Reddit gets indexed by Google within days. That means a single bounty campaign builds AI citation signal and SEO backlink signal simultaneously. For most brands, the SEO lift is a side effect they didn't expect. The creator pieces targeting your priority queries end up ranking on Google for those same queries, sending organic traffic on top of the AI-referred traffic. One brief. Two channels. Both building over time.

The citation economy is what AI search runs on. In SEO, the currency was backlinks, the more authoritative sites linked to you, the higher you ranked. In AI search, the currency is citations, the more credible, consistent, third-party sources mention your brand in the right context, the more likely AI is to name you in the answer. The shift matters because the rules changed completely. You can have a perfect website, a strong domain authority, and years of SEO investment, and still be invisible to AI. That's because AI doesn't rank pages. It builds answers from signals. Brands that understand the citation economy early are building a moat that compounds. Brands that don't are watching their competitors get recommended instead.

Every citation is tracked end-to-end in the Scribble dashboard. You can see which creator published what, on which platform, on which date, and which AI models are now citing that piece in their answers. We run the actual queries live and show you the model's response with your brand in it. Not a report. Not a projection. The answer, pulled in real time, with your brand named in it. That's the proof of work. You can also see citation movement over time, your share going from 2% to 9% to 20% isn't an estimate, it's the live query result averaged across your tracked basket. If a campaign moved your number, you'll see exactly which pieces did it.

We run your tracked queries live against each AI model's API, ChatGPT, Gemini, Perplexity, Copilot, and Grok. We then parse every response for two things. A mention is when the model names your brand in its answer text. A citation is when the model explicitly references a source that points to your domain or to creator content about you, the link or source attribution that appears below the answer in models like Perplexity. Your citation score is calculated by running every query in your tracked basket, counting how many responses include a mention or citation, and expressing that as a percentage. Do it across five models and you get a full picture of where you're winning, where you're invisible, and which model is your biggest gap.