# Mosaic > Mosaic is the AI-native omnichannel marketing intelligence platform: paid media, organic search, social and AI answer visibility in one warehouse, every creative read and scored by Prism, and multi-model recommendations your team approves. Mosaic is built by Archon Agents (https://archonagents.ai). Contact: archon@bmistudios.com. ## What it does - Unifies paid media, measurement, organic search, social and AI answer engine data in one warehouse, refreshed every night. - Prism creative intelligence: stores every image, video and carousel, counts copies of the same ad across ad sets as one creative, watches what is in each one, and gives it a 0 to 100 Prism Score against similar ads (Stop and Hold for video, Click, Convert, Fit) with the reasons in points, findings on what works and ideas to test next. It ranks attention and clicks on ads it was not trained on; it does not forecast bookings or revenue. - Multi-model orchestration: a rules engine computes every number in SQL, a classifier model tags campaigns, ads, queries and AI answers, and an analyst model explains. Models abstain below a confidence threshold and are benchmarked nightly. - Recommendations your team approves or declines; approved moves go out in a weekly report. - Knowledge base and institutional memory: brand facts, naming and glossary every model reads first, a decision log with reasons, nightly snapshots and a creative archive. - Read-only MCP connectors so Claude, ChatGPT and local models can query the warehouse. - Client data is never sold, distributed or used to train AI models. ## Connects to - Paid media: Google Ads, Meta Ads, TikTok Ads, Microsoft Ads, The Trade Desk, Campaign Manager 360 - Measurement: Google Analytics 4, Floodlight conversions - Organic search: Google Search Console, Landing pages (GA4) - Social: Instagram, Facebook, TikTok, Sprout Social - AI stack: Claude, ChatGPT, Local models, Jev by TypeSafe, Peec AI, Semrush - AI answer engines: ChatGPT, Gemini, Google AI Overviews, Google AI Mode, Perplexity, Copilot ## FAQ ### What is Mosaic? Mosaic (Marketing Operations, Spend and Analytics Intelligence Core) is the AI-native omnichannel marketing intelligence platform. It brings paid media, organic search, social and AI answer visibility into one warehouse every night, scores every creative, and turns the data into specific recommendations your team approves. ### Which platforms does Mosaic connect to? Google Ads, Meta, TikTok, Microsoft Ads, The Trade Desk and Campaign Manager 360 for paid media; Google Analytics 4 and Floodlight for measurement; Google Search Console and Semrush for organic search; Instagram, Facebook and TikTok through Sprout Social for social; and Peec AI for ChatGPT, Gemini, Google AI Overviews, Google AI Mode, Perplexity and Copilot. ### How does Mosaic use AI without making things up? Every number is computed in the warehouse with SQL, never by a language model. Specialist models only label, rank and explain. When a model is not confident it abstains and the item is marked for review, and every AI task is scored nightly against hand-labelled gold sets. ### What is multi-model orchestration? Mosaic routes each job to the model built for it instead of asking one model to do everything. A rules engine computes every number in SQL, a classifier model tags campaigns, ads, search queries and AI answers, an analyst model explains and answers questions, and Prism watches and scores every creative. The orchestration layer combines their work, checks it, and hands your team a recommendation to approve. ### What is the Mosaic knowledge base? The knowledge base holds your brand facts, naming conventions, funnel definitions and glossary. Every model reads it before it labels, explains or recommends anything, so answers use your language and your rules, and new team members inherit the same context on day one. ### What is Prism? Prism is Mosaic's creative intelligence agent, built to help your team make better creative decisions. It starts from each creative's DNA (how well it hooks, holds and engages) and adds what is in the ad, from the opening seconds to the people, offer, text and sound, then compares it fairly with similar ads. Every creative gets a 0 to 100 Prism Score with the reasons in points, plus findings on what works and ideas for what to test next. Scores are tested on ads Prism was not trained on, where they rank attention and clicks well. Prism does not forecast bookings or revenue; those stay measured in the warehouse. ### Can I use Mosaic from Claude, ChatGPT or a local model? Yes. Mosaic exposes read-only tools through the Model Context Protocol (MCP), so Claude, ChatGPT and MCP-compatible local models can ask questions, pull reports and read recommendations. Access is switched on per person and every call is logged. ### Is our data used to train AI models? No. Client data is never sold, distributed or used to train AI models. AI features have read-only access and cannot change your data or your ad accounts. ### What does Mosaic remember? Your brand facts, naming conventions, funnel definitions and glossary, every creative you have run, and every recommendation with the decision your team made and why. That memory stays when people change roles. ### How is Mosaic different from a dashboard or a data connector? Connectors move data and dashboards display it. Mosaic also interprets it: it reconciles definitions across platforms, labels and scores everything, recommends the next move with the evidence behind it, and records what your team decided. ## Links - [Home](https://mosaicsignal.ai/) - [Sign in](https://app.internal-hub.com/login) - [Privacy](https://app.internal-hub.com/privacy) - [Terms](https://app.internal-hub.com/terms)