PermitReady: Your AI co-pilot from business idea to legally registered in days

Problem Statement

MSMEs make up approximately 99% of all registered businesses in the Philippines, according to PSA and DTI. Millions more operate informally, partly because registration feels intimidating. A new sole proprietor typically has to register the business name with DTI (BNRS), get a barangay clearance, secure a Mayor’s/business permit through the LGU’s BOSS, register with BIR (ORUS, Form 1901, books of accounts, invoices), and often enroll with SSS, PhilHealth, and Pag-IBIG. Each agency has different portals, forms, and jargon, and the steps depend on each other. Many first-time owners pay fixers or give up. Because they remain informal, they miss out on loans, government programs, and online marketplaces that require a TIN or permit.

Proposed Solution

PermitPilot is a conversational agent on web and Messenger. The owner describes their business in plain Taglish (“magbebenta ako ng ulam online, sa bahay lang sa Pasig”). The agent determines the business type, the likely PSIC/line-of-business code, the BIR tax type (e.g., the 8% option or graduated rates, with explanations), and the exact sequence of steps for that specific LGU. It asks only the necessary questions once, then generates pre-filled drafts of DTI, barangay, LGU unified application, and BIR 1901 forms for the owner to review and submit on official portals. A status tracker shows each step (submitted, for payment, released), sends SMS reminders for renewals and BIR deadlines, and lets the owner upload receipts. The vision module reads the issued certificate to auto-advance the next step.

Target Market

• Primary users: First-time micro-entrepreneurs, online sellers (pressured by e-commerce and BIR registration requirements), OFW families starting businesses, and youth entrepreneurs.
• Secondary / B2G / B2B: DTI Negosyo Centers, LGU BPLOs, and BOSS offices (white-label tool), banks and microfinance institutions (onboarding funnel), and e-commerce platforms (seller onboarding).
• Size: Approximately 1 million+ registered MSMEs and a much larger informal sector. DTI registers hundreds of thousands of new business names each year.

How AI Is Used

• LLM agent with planning and tool use: Reasons over the owner’s situation and builds a personalized multi-agency roadmap. Static checklists cannot cover the combinations of business type, location, and tax regime.
• RAG over official guidelines: Grounds answers in DTI, BIR (Revenue Regulations, RMCs), and LGU charters so the tool doesn’t invent rules.
• Structured extraction and form filling: Converts conversation into validated form fields, which is the main time-saver.
• Vision OCR: Reads issued DTI certificates, barangay clearances, and permits to update status and carry data forward (e.g., business name number into BIR forms).
• Classification: Suggests PSIC codes and BIR tax types with plain-language explanations.

PLDT / Smart / ePLDT / Rokid Angle

• PLDT Enterprise SME base: A natural value-add for PLDT/Smart business broadband and SME bundles (“Get connected and get registered”).
• Smart SMS: Renewal and tax deadline reminders (e.g., the January business permit renewal and quarterly BIR filings).
• ePLDT cloud: Hosts the agent and secure document vault, with an option to white-label for LGUs and DTI.
• Rokid AR (stretch): Negosyo Center staff see a client’s PermitPilot progress overlaid during walk-in assistance.

Demo Scenario (3 minutes for judges)

1. Meet Joy, 23, who sells homemade leche flan on Facebook and got a message saying she needs a permit to keep selling on a platform.
2. She chats in Taglish. In 60 seconds PermitPilot builds her roadmap: DTI → Barangay → Pasig BOSS → BIR (8% option explained simply).
3. One click generates her pre-filled DTI and BIR 1901 drafts, with fields highlighted for her review.
4. She uploads a photo of her (mock) DTI certificate. OCR reads the BN number and the tracker advances, auto-filling it into the LGU form.
5. The timeline shows the remaining steps with estimated days and fees, and a mock SMS reminder arrives.
6. Close: “Formalizing 1% more micro-businesses means more credit access, more tax base, and fewer fixers.”

30-Hour Build Plan

Suggested stack: Next.js, LangGraph or a simple function-calling agent with Gemini/GPT-class models, pgvector, pdf-lib for form filling, PaddleOCR, Supabase, and an SMS gateway (Smart in production). Public data / APIs: DTI BNRS public guides, BIR forms and Revenue Regulations (bir.gov.ph), LGU BOSS and Citizen’s Charters, PSA PSIC classification, PSA OpenSTAT MSME statistics, and data.gov.ph.

Risks & Privacy Notes

• Regulatory accuracy: Rules change, so the tool cites sources, versions the knowledge base, and shows a disclaimer that it is guidance and not legal or tax advice. Final submission happens on official portals.
• No official APIs: The tool pre-fills drafts instead of auto-submitting and tracks status from user updates and OCR. Agency integration is a post-hackathon goal.
• Data Privacy Act 2012: TINs and IDs are encrypted, with explicit consent, data minimization, and user-controlled deletion.
• Clause 24 compliance: Only mock personas and synthetic documents are used during the build. No real TINs or personal data go into AI tools.

AI Use Disclosure (draft)

We use an LLM (e.g., Gemini) as a planning agent and RAG assistant over public DTI, BIR, and LGU guidelines, and an OCR model for reading certificates. AI coding assistants helped with UI and PDF-filling code, and mock personas and documents were AI-generated. Team members verified all regulatory content against official sources and reviewed all code. No personal or confidential data was entered into AI tools.

Rubric Scoring