Scan ad copy and landing pages against Meta, Google, TikTok, and Taboola policies. Get safe rewrites that keep your hook.
Ad headlines, body copy, CTAs — or a full landing page URL. We handle text or HTML, structured or messy.
Your input is embedded and matched against Meta, Google, and Taboola policies via pgvector. Top-8 most relevant rules reach the LLM.
A critic agent reads your copy through the lens of real platform policy — flagging exact phrases, classifying severity, returning a risk score.
8 distinct copywriting techniques. Empowerment framing, curiosity hooks, social proof. Safe — but never sterile.
Paste raw copy or hand over a URL — both flow through the same critic + optimizer pipeline.
Paste your headline, body, and CTA. We score it against every platform policy in seconds.
Drop a URL. We scrape the page, parse the content, and flag bait-and-switch patterns automatically.
Stop repeating yourself. Scan once with tadan, and we will check your copy against every platform policy in one pass.
Stop losing time, money, and accounts to compliance. One scan, every platform.
What used to take a day of cross-referencing four platform docs now takes six seconds. Paste your copy, get a verdict.
RAG over curated policy rules from Meta, Google, Taboola, and TikTok. Not vibes. Not LLM guessing.
Eight copywriting techniques. Empowerment framing, curiosity hooks, social proof. Safe variants that still convert.
One scan covers Meta, Google, and Taboola. Every rule. Every platform. Every variant.
Yes. tadan is open source under the GNU AGPL-3.0 license. You can browse the code, submit issues, and contribute at github.com/sudoeren/tadan. Self-hosting is straightforward — the README covers the full setup.
Meta Ads (Facebook + Instagram), Google Ads, TikTok, and Taboola. Outbrain, X Ads, and LinkedIn are on the roadmap. The policy database is modular, so adding a new network is a single new file plus a one-line addition to the policy map.
A 0–100 score where 0–25 is clean, 26–60 needs review, 61–85 is risky, and 86+ is bannable. The score is the maximum across all selected platforms.
Most scans finish in 10-30 seconds end-to-end. The critic agent runs first to find violations, then the optimizer agent rewrites them in parallel with the database save. You'll see four live stages: loading policies, reading content, analyzing, and rewriting.
Our optimizer agent uses 8 distinct copywriting techniques — empowerment framing, curiosity hooks, social proof, specificity, and more. Each variant is split into headline, body, and CTA so you can drop them straight into your ad platform.
Yes. Drop a URL, we scrape with Cheerio, format the content, and run the same pipeline. We also flag metadata issues like missing privacy policies and bait-and-switch headlines.
When your ad is clean, the critic agent lists the specific things it did well — quoting the actual headline, CTA, or claim. It's not a generic checklist. If nothing specific comes back, the section stays hidden.
Yes. Open any scan from history to view it, or use the Select button to pick one or many scans and delete them in bulk. Deletion is permanent and also wipes the linked violations and rewrites.
Your ad is embedded with OpenAI's text-embedding-3-small and matched against policy rules stored as pgvector embeddings. Only the most relevant rules are passed to the critic LLM, so the model focuses on what actually matters for your copy.
No. Your ad copy, landing pages, and scan history are never used for training. Embeddings live in your own database. Since tadan is open source, you can verify this yourself — or self-host and keep everything entirely on your own infrastructure.
Yes. The repo includes a Docker Compose setup for PostgreSQL + pgvector, a predev script that auto-initializes the database, and a standalone Dockerfile for production. All you need is an OpenRouter API key. See the README for the full guide.
Yes. You can scan ad copy and landing pages, review results, and access your scan history on your phone, tablet, or desktop.
Yes. tadan is open source under the GNU AGPL-3.0 license. You can browse the code, submit issues, and contribute at github.com/sudoeren/tadan. Self-hosting is straightforward — the README covers the full setup.
