MentionScope
A Windows desktop app that asks AI assistants the questions your customers ask, then measures how often, and how highly, they recommend your brand.
Preview only. This one isn't open for use.
What it is
More people now ask an AI assistant for a recommendation before they search the web. If someone asks for the best project management software, which brands does the assistant name, in what order, and with what tone? MentionScope answers that with data instead of guesses.
It runs real searches against several AI providers, turns every free-text answer into structured results, and stores each run so you can watch the trend over time. It tracks your brand and your competitors together, so you can compare yourself per model and per question.
Walkthrough
- Add your brandsIn the Brands tab, add your own brand and the competitors you want to compare against. Every run looks at all of them at once.
- Write your promptsIn the Prompts tab, add the questions a buyer might actually ask an AI about your category.
- Connect your modelsIn the Models tab, paste in your API keys, hit Test, and enable the providers you want. Keys are stored encrypted and the app only ever shows a masked hint.
- Run a searchClick Run search. Prompts go out to every enabled provider in parallel, with progress showing live. A failure on one request never stops the rest, and a cancelled run can be resumed later.
- Read the dashboardSee your visibility percentage, average rank, mentions and a leaderboard of every brand, with trend arrows against the previous run.
- Dig into the reportsReports covers cost, visibility and how each model performed. Every raw answer is kept, so you can open any one and read exactly what the AI said.
Under the hood
- One consistent reader. Instead of parsing each provider's writing style with patterns, every raw answer goes to one model using strict structured output, validated again in code. For each brand it records whether it was mentioned, its rank, the sentiment and how strongly it was recommended.
- A careful run engine. It limits concurrency overall and per provider, backs off with jitter when a provider pushes back, classifies errors, and keeps going when one request fails.
- Easy to extend. Adding a new OpenAI-compatible provider is a single entry in a registry.
- Durable history. Every task stores its raw response, latency, token use and estimated cost, so trend analysis and cost reports come for free.
The desktop app
MentionScope also ships as a standalone Windows installer built with Electron. It starts its own server in the background, creates a private database and encryption key for each user, and opens like any normal program. There is no terminal and no config file. Customers add their own API keys after installing.
The preview above is a rebuild of the dashboard with invented brands and numbers.