I built this in a day. Not because it was easy, but because I wanted to prove something: Indian founders don't need another complicated dashboard. They need someone to tell them, in plain English, what's broken and what to fix first.
Last weekend, I sat down and built CompliScore – a free 60-second compliance health scanner for Indian startups. You type your company name, and it gives you:
A Compliance Score (out of 100)
A list of pending GST & MCA filings
A penalty risk level (Low / Medium / High)
An AI‑generated action plan written in simple language
🔗 Live Demo: compliscore-nu.vercel.app
Why I Built This
I'm Neha, a software engineer from Mumbai. I've been deep in the compliance automation space for months, building DigiNav AI – an autonomous agent that will eventually handle actual GST filings, incorporation, and audits end‑to‑end.
But the big product takes weeks to build. I needed something today that could help founders right now and also serve as a live portfolio piece. So I stripped everything down to the single most useful action: Show me my risks.
I've also contributed 150+ PRs to core AI agent frameworks (Mastra, LlamaIndexTS, LiteLLM, PostHog) – so I understand both the compliance domain and the AI orchestration under the hood.
What I Learned While Building It
- Founders don't care about compliance data. They care about consequences. Most tax-tech tools dump a table of due dates. That's useless. What a founder actually wants to know is:
Am I about to get a penalty?
How much will it cost?
What's the ONE thing I should do right now?
I designed the scanner to answer exactly that. The score and risk badge give an instant emotional signal (green = relief, red = danger), and the AI‑written action plan gives a clear next step in 150 words.
2. Speed builds trust.
The scanner uses mock data for now (real GST/MCA integration is in progress), but the UI flow and the AI logic are production‑ready. I built it in <8 hours using Next.js, Tailwind, shadcn/ui, and Groq for the AI summary. When you ship fast and share publicly, people notice. Within 48 hours of launching, I got an inbound message from a VC‑backed founder building in the exact same space.
3. Personal projects are the strongest resume
.
My GitHub account was flagged due to a 2FA loss, and I lost direct access to my old profile (150+ PRs). But this scanner, built in a day and deployed publicly, has done more for my credibility this week than my entire contribution graph did in a year. Recruiters and founders want to see live proof, not just a heatmap.
Tech Stack (for the curious)
Frontend: Next.js 15 (App Router), Tailwind CSS, shadcn/ui, Framer Motion
AI Layer: Groq API (Llama 3.3 70B) – generates the plain‑English action plan
Data Engine: Static mock data with a scoring algorithm that calculates compliance health based on overdue filings, missing returns, and pending notices
Deployment: Vercel (free tier, instant deploys)
What's Next
I'm currently building the full autonomous agent (DigiNav AI) that will connect to actual GSTN, MCA, and DigiLocker APIs and execute real filings. The scanner is the first step – a free tool that demonstrates the UX I want the full product to have.
If you're a founder and want a real, manual compliance health scan for your business, I offer those for ₹5,000 – delivered in 2 hours. DM me here or on Twitter/X / LinkedIn.
If you're building in the tax‑tech or AI agent space, I'd love to connect and compare notes. Drop a comment or reach out directly.
My links:
Portfolio: https://neha-portfoliooo.vercel.app/
GitHub: https://github.com/nehaprasad-dev
Live Scanner: https://compliscore-nu.vercel.app/
Top comments (2)
The "founders don't need another dashboard, they need to be told what's broken and what to fix first" insight is the whole game, ranked action beats raw data every time. Shipping it in 24 hours is exactly the right move too, because the only way to learn whether founders actually want a compliance score is to put a live one in front of them, not to spec it for a month. The interesting hard part with anything compliance-flavored is trust: the moment the AI action plan cites a wrong filing or a penalty risk that's off, you lose the user, so the score is only as good as the data behind it and the guardrails on the generation. I'd grade the AI plan against the structured GST/MCA facts before showing it, refuse rather than guess when data's missing. That verify-before-ship discipline is what I build into Moonshift. What's your source of truth for the filings data, and how do you keep the AI from confidently inventing a deadline?
I really appreciate this you've nailed the exact tension I've been thinking about. "Verify before ship" is the right discipline, and you're right that one wrong penalty estimate or outdated deadline can destroy trust instantly.
Right now, CompliScore runs on mock data. I built it in a day specifically to test whether founders even care about a compliance score before committing to a full production backend. The response has been strong enough that I'm now mapping out the real data pipeline: pulling from public GST/MCA portals, cross-referencing filing history, and using a deterministic scoring layer that feeds the AI structured facts, not raw text. The AI only gets to write the action plan after the numbers are locked it can't invent a deadline because the deadline is already hard-coded in the structured payload.
Longer-term, for DigiNav (the autonomous agent I'm building on top of this), the plan is a hybrid gateway: official APIs (GSTN, MCA) where available and browser automation with retry logic for the portals that have no APIs. The agent will also log every data source, decision, and change, so there's a full audit trail.
Would love to hear how you're approaching the trust/verification layer in Moonshift sounds like we're solving similar problems from different angles. Happy to connect further if you're open to it