Every day, someone writes a post that starts with “Can anyone recommend…” or “Is there anything simpler than…”. Some of those posts describe exactly the problem your product solves.
They are public. They are also easy to miss. They sit on Reddit, X, LinkedIn, YouTube, TikTok and Instagram, between thousands of posts that use the same words and mean something else. Whoever finds the question first can give the useful answer.
SignalScout finds those conversations for you.
Keyword alerts find words, not people
The usual tool for this is a keyword alert. If you sell an invoicing app, you set an alert for “invoice”. It then sends you template giveaways, accounting tips, job posts and lifetime deals. Somewhere in that list is the one person who asks for a tool that chases late payments.
A keyword alert cannot tell those posts apart, because it only matches text. You read all of them to find the one.
What SignalScout does instead
You describe your product, the people who buy it, and the problem it solves. SignalScout then searches the platforms you choose and reads each conversation with an AI model. It scores three things:
- Problem fit: does the post describe the problem you solve?
- Customer fit: does the person look like your customer?
- Intent: are they looking for an answer now?
Posts that score well go to an inbox, ranked by score and age. Each one keeps the source text and the reasons for its score, so you can check the verdict yourself instead of trusting a number.
Before the AI model reads a post, three cheaper filters remove the posts that cannot match. On a broad search, most posts stop there, so the model is paid to read only the posts that might matter.
You can read the inbox, or get the matches by email digest, immediate email, signed webhook or CSV export.
It does not post for you
When a conversation is worth answering, SignalScout can draft a reply in a voice you define. You read it, edit it and post it yourself. Nothing is posted from SignalScout.
This is deliberate. A reply in a public thread should come from a person who has read the thread.
What it is not
SignalScout has no sentiment charts, no share-of-voice reports, no word clouds, no competitor analytics, no social publishing and no CRM. Those are good products. They are not this one.
The scope is one sentence: find conversations with intent.
Two ways to run it
SignalScout Cloud is the hosted version. Social data and AI usage are included in the plan, so you add no API keys. The trial is seven days with no credit card: one project, four search credits a day on Reddit, X, YouTube and TikTok, and twenty reply drafts. After the trial, plans start at $19 a month. The pricing page has the full comparison.
Self-hosted is the same application on your own server. The complete source is on GitHub: the monitoring pipeline, AI scoring, inbox, alerts and reply drafts. It runs with Docker, and you use your own data-provider and AI-model keys. You pay those providers directly. As a rough guide, fetching a Reddit post costs $0.0003 to $0.0015, and scoring it on a cheap model costs about $0.001.
Cloud charges for not running a server. It does not unlock features that the self-hosted version lacks.
Try it
Start a seven-day Cloud trial, or follow the self-hosting guide. The first-match walkthrough takes about ten minutes either way.
On this blog, we will write about what we learn while we build SignalScout in the open: how we score posts, what each platform costs, and the decisions behind the product.