01
Paste or upload feedback
Drop in raw customer reviews from anywhere — no schema, no setup — and FeedLens takes it from there.
An AI SaaS that turns raw customer reviews into decisions — sentiment, recurring themes and the concrete action items worth a team's time, in one calm dashboard. Designed, built and owned end to end.
Customer feedback piles up faster than anyone can read it — scattered across stores, forms and support threads. Reading it one review at a time doesn't scale, and a gut-feel summary misses what to actually do.
Teams don't need more dashboards of raw text — they need the signal: how people feel, which themes keep recurring, and the specific next step. FeedLens was built to turn the pile into a short, honest list of decisions.
A full AI SaaS: paste or upload feedback, let the model do the reading, and get back sentiment, themes and action items — with the auth, plans and billing of a real product around it.
Drop in raw customer reviews from anywhere — no schema, no setup — and FeedLens takes it from there.
Each review gets a sentiment (pos/neu/neg), a 1–5 score, up to 3 recurring themes, and one concrete action item.
Sentiment over time, top themes and an action-item queue — the signal, not the noise, in one calm view.
Auth, workspaces, a metered free tier and a Pro plan with checkout — the full SaaS surface.
Reviews are analysed by gpt-oss-120b via Groq under a strict-JSON contract, so each one comes back as clean, typed data — sentiment, score, up to three themes and a specific action — same length and order as the input, no free-text drift.
Supabase Auth with a server-side session (@supabase/ssr), so every query runs as the signed-in user across workspaces.
A metered free tier (50 analyses/month), per-user rate limiting and Pro gating keep the AI spend and abuse in check.
Checkout via Xendit with a signature-verified webhook — a PAID/SETTLED invoice flips the workspace to Pro. (Billing is wired end-to-end and currently runs in test mode.)
Recharts-driven sentiment-over-time and theme breakdowns turn the AI output into something a team can actually read at a glance.
Next.js 16 App Router, a real login, settings, theming, and privacy/terms pages — built like a product, not a demo.
I build AI into real workflows and run a SaaS of my own. New product or an update to an existing one — let's talk.