GreenLens AI

Inspiration

GreenLens AI was inspired by a real conversation with a client. He told me he needed an ESG report to help secure funding for his growing business, but consultants were charging thousands of dollars for something he urgently needed and simply could not afford at that stage. As a GenAI developer, that moment made the problem feel obvious: small and medium-sized businesses need credible ESG reporting, funding guidance, and compliance support, but the current process is too expensive, too slow, and too inaccessible. GreenLens AI was built to make that process fast, affordable, and practical.

What it does

GreenLens AI turns a company’s financial and operational data into a lender-ready ESG intelligence report in minutes. Users upload a transaction CSV and can optionally upload supporting documents such as invoices, utility bills, and receipts. The platform then analyzes emissions across Scope 1, 2, and 3, calculates an ESG score, checks compliance readiness against Canadian frameworks, surfaces matched grants and funding opportunities, and generates a polished report and dashboard.

What makes it especially useful is that it does more than just summarize data. It also performs supporting-document assurance and fraud detection by matching uploaded documents against ledger activity, flagging duplicates, mismatches, suspicious patterns, and transaction anomalies. The result is a report that is not only informative, but also more trustworthy for a financial institution or stakeholder reviewing it.

How we built it

We built GreenLens AI as a full-stack application with a Next.js frontend and a FastAPI backend. The frontend handles intake, live analysis progress, dashboard views, a full ESG report, and a separate fraud detection experience. The backend processes uploaded CSVs and PDFs, normalizes transaction data, classifies emissions into Scope 1, 2, and 3, calculates benchmarks and ESG scores, retrieves compliance and grant context, and generates the final report.

We combined deterministic analytics with GenAI. The numerical outputs, fraud checks, emissions calculations, and compliance matching are rule-based so the results stay grounded and explainable. Then we use an LLM to convert those verified outputs into a clean narrative report and report copilot experience. We also added fallback logic, document parsing, fraud heuristics, and end-to-end status tracking so the demo works reliably in real time.

Challenges we ran into

One challenge was making the product feel credible, not just flashy. ESG reports involve real numbers, compliance expectations, and financial decisions, so we could not rely only on generative text. We had to make sure the underlying calculations, fraud checks, and report structure were deterministic and consistent.

Another challenge was keeping the demo stable while adding more ambitious features. We ran into UI crashes after analysis completed, had to normalize report data across frontend and backend, fixed session and form-state issues, and made sure the app worked both with and without supporting documents. We also had to handle LLM reliability, print/PDF formatting, and fraud workflows without breaking the core ESG experience.

Accomplishments that we're proud of

We’re proud that GreenLens AI takes a problem that usually costs thousands of dollars and compresses it into a much faster, more accessible workflow for small businesses. The platform does not just generate a report, it creates a complete experience: intake, analysis, dashboard, compliance context, funding matches, fraud detection, and a polished report.

We’re also proud of the fraud detection layer. Instead of treating uploaded documents as passive attachments, GreenLens AI actively checks them against the ledger and surfaces meaningful issues like duplicates, mismatches, weak evidence coverage, and anomalous transaction patterns. That makes the demo more compelling and makes the output feel more institution-ready.

What we learned

We learned that the strongest GenAI products are not pure generation tools, they are systems that combine reliable structured analysis with AI-generated explanation. The trust comes from deterministic data processing; the usability comes from turning that into something understandable and actionable.

We also learned how important product polish is. Small issues like prefilled forms, broken report tabs, missing defaults, or messy PDF exports can undermine confidence, especially in a financial or compliance context. Building a strong demo meant paying as much attention to UX reliability as to the intelligence behind the scenes.

What's next for GreenLens AI

Next, we want to make GreenLens AI even more useful for real businesses and financial institutions. That includes improving export quality for truly lender-ready PDFs, expanding support for more ESG and regulatory frameworks, adding richer OCR and document ingestion, and integrating directly with accounting platforms so the workflow becomes seamless.

We also want to grow the platform from a reporting tool into a continuous ESG intelligence system. That means ongoing monitoring, multi-period benchmarking, stronger fraud and assurance workflows, collaborative review tools, and clearer recommendations tied directly to funding eligibility and business growth.

AI Use

Approximately 60% of the code was generated or assisted by AI tools (e.g., ChatGPT / code copilots) and then reviewed, modified, and integrated manually.