Roadmap

• Making the first paid subscription verifiable end to end. Payments are handled by Paddle, and the path from “payment received” to “plan switched on” is the one link in the chain we have not yet watched complete against real money. Until it has, we are not promoting the product, because a checkout that takes a payment and grants nothing is the worst failure this software can have. If you buy during this period and anything about your access is wrong, write to us and it will be fixed immediately.
• These explanation pages. The score is only worth something if you can see how it is produced, so methodology, security and the API reference are being written and corrected against the code, not against the plan.

• More engines in every scan. The measurement code already knows about ChatGPT, Claude, Gemini, Perplexity, Qwen, Kimi, DeepSeek, Doubao and Yuanbao; today two of them are switched on. Each additional engine is a paid API call multiplied by every question in every scan, so engines get added as revenue allows rather than all at once and at a loss. This is the single biggest improvement to the product, which is why it is first on the list.
• Email alerts when your score moves. The daily monitoring already detects a real change and separates it from noise — which is the hard part. Nothing emails you yet, so today you have to look. This also means we need a deliverable sending domain, which is the same work as moving support off a personal mailbox.
• Off-site backups and a restore we have actually run. The database currently lives on one server. A backup that has never been restored is a belief, not a backup, so the item includes the rehearsal.
• A tighter Content-Security-Policy. The current policy still allows inline and eval scripts. Moving to nonce-based script loading is a real hardening step, and it is listed in the open gaps on the security page.

Not scheduled, no promise implied, listed because these are the most common requests and you deserve to know the answer is "not yet" rather than "never".

• Two-factor authentication and an audit log of sign-ins.
• More industry categories in the public AI visibility index.
• CSV and spreadsheet export of a scan's raw answers.
• Additional seats on one account, so a team can share a project.
• Per-key permissions for the API, instead of one key per account.
• Watchlists: tracking a competitor whose domain you do not own.

A roadmap that only grows is a sign nobody is saying no. These are the things we are not going to build, with the reason:

• Writing your content. The product tells you what is missing and where; the words have to be yours, because the point is to be found for what you actually do.
• A chat interface that answers questions about the score. The report is already the answer, and a chat window over a 0–100 number is a way to make it feel sophisticated while making it less checkable.
• An "AI mentions" vanity counter. A big number that goes up is easy to sell and easy to fake. The score, its interval and the raw answers are harder to sell and possible to verify, and we would rather have the second one.
• Monthly reports you have to log in for. Making a customer come back to a dashboard to read their own data is a retention trick, not a feature. When reports can be emailed, they will be emailed.

Write to us and say what you need and why. A request from someone who is paying for the product outranks one from a competitor-analysis spreadsheet every time — and a request from a customer outranks anything on this page, including the items marked as being worked on now.