LadyBugs
Every business in Azerbaijan that pays bills - from small accounting teams and accountants to CFOs, procurement managers, and business owners. The Problem - Businesses lose thousands of AZN every month to a silent budget killer: overbilling. Suppliers send invoices (Qaimə) with slightly inflated prices, stealthy price hikes, or unapproved charges that don't match the original contract (Müqavilə). Checking line items by hand is tedious, humanly exhausting, and virtually impossible to catch at scale. UcotAI - Snap a photo of an invoice or contract, and UcotAI does the rest. Our AI scans the document, reads every single line item, compares it against contract terms, and immediately flags price markups, tax errors, and hidden overcharges before the money leaves the company's bank account.
Preliminary score
- Value for the user16.5 / 25
A specific problem and a clear outcome from solving it.
- Prototype and use of AI12.5 / 30
A working core scenario, and what AI actually contributes to the result.
- Quality testing4.5 / 20
Tests, examples of failures, and comparison with the current approach.
- Feasibility5.5 / 15
Data requirements, running costs and a clear next step.
- Originality5.0 / 10
A meaningful difference from common solutions.
What the judges wrote
- Claude Opus 5.5Preliminary43/100
- Strengths
- Your written submission names a concrete, costly pain: suppliers' invoices (Qaimə) drifting from the prices agreed in the contract (Müqavilə), caught before payment. Your architecture is sensible: Gemini Vision extracts VÖEN, ƏDV 18% totals and line items from photos, and a deterministic reconciliation step computes the exact price gap in AZN. That keeps the money-critical comparison out of the LLM. Grounding the tool in local document formats makes it more relevant than a generic accounting app.
- To improve
- Your deck pitches a broad accounting SaaS (inventory, P&L, forecasts) rather than the invoice-vs-contract audit you actually built, so align the story and the demo around the one workflow. Your testing covers only one invoice and one contract, with no reported results. Run a set of document pairs with seeded markups and show how many discrepancies were caught or missed, the OCR errors on messy photos, and the time against manual checking. Replace unsupported figures such as '~99% error elimination', '65% of SMBs' and '18-month runway secured' with evidence, add Gemini API cost per document, and put real setup instructions in the README.
- GPT-6 AstraPreliminary45/100
- Strengths
- Your invoice-versus-contract audit targets a concrete pre-payment problem for Azerbaijani accounting teams, with discrepancies expressed in AZN. Your separation of Gemini document extraction from deterministic price and 18% ƏDV checks is a sensible design for financial auditing. Your initial testing identifies specific local fields—VÖEN, contract references and line items—rather than relying solely on generic accuracy claims.
- To improve
- Show a reproducible invoice/contract pair with extracted fields, matched items, expected discrepancies and actual audit output; the materials describe the workflow but do not demonstrate its result. Expand beyond one pair to test blurry photos, ambiguous product names, quantity mismatches and tax errors, reporting failures and measured comparisons with manual checking. Document data-access permissions, financial-data safeguards and processing costs per document rather than subscription prices alone. Align your broad accounting-platform deck with the submitted overbilling workflow and define a focused pilot as the next step.
The team
Nigar Nazarli
data
Leyla KhudiyevaLead
product
Gulnar Hasanli
developer