Acre — Privacy-Preserving Underwriting Framework for Gig Workers

Configurable. Regulatory-Safe. Zero DPDP Liability.

Onboard India's 8M+ credit-invisible gig workers safely under RBI/DPDP norms.
Configure your policy. Zero raw PII. Immutable audit trail.

Problem · Solution · How it works · Get started · Business model

The Problem

RBI 2025 Blocked You. DPDP Threatens Your Balance Sheet.

RBI Digital Lending Directions (2025) banned SMS, contact lists, location tracking, and device fingerprinting for underwriting. DPDP Act (2023) threatens ₹250 crore penalties for raw PII misuse.

Your situation:

You originated 301 lakh sub-₹50k accounts in Q1 FY26 (NBFC-fintech data)
You want to lend to India's 8M+ gig workers — the exact segment
Your old alternative-data playbook is now criminal and expensive
Account Aggregator only covers bank flows — misses platform tenure, ratings, completion rates that predict gig-worker repayment
Result: ~40% approval rate on gig workers, even good credit risks, because you're missing signals

The Cost of Silence

Challenge	Your Cost
Manual thin-file underwriting	₹800–₹1,200 per applicant
DPDP compliance team	+3–5 people, ₹50–100L/year
Market loss	8M gig workers you can't safely touch
Regulatory audit prep	₹20–50L per inspection cycle

The Solution

Acre is not a credit score. It is a configurable, privacy-preserving underwriting framework sold to lenders.

What you get:

You define the policy — Select which proof modules (income, tenure, rating, identity) matter to YOUR risk appetite. Set thresholds. Adjust weights. No black box.

Workers verify locally — QR scan → Login to Uber/Swiggy → ZK proof generated on their phone → Only YOUR configured signals revealed. Zero custody of raw data.

Blue Score computed — Explainable 300-900 scorecard → Blue Prime / Blue Plus / Blue Basic tiers. You know exactly what drives each worker's score.

Proof stored immutably — Consent artifact + proof hash on Algorand. Immutable audit trail for RBI/DPDP inspections. Zero raw PII on your servers.

You query in real-time — get_eligibility(worker_wallet) → Instant credit tier, limit, proof timestamp. Your compliance team sleeps.

What Changes for You

Metric	Before	With Acre
Cost per verification	₹950 (manual)	₹55 (API)
Approval rate on gig workers	<40%	70%+ (3x lift)
Compliance cost	+₹50–100L/year	–70% (audit trail built-in)
Market accessible	0 gig workers safely	8M+ gig workers
Raw data custody	You store it (DPDP risk)	Zero; only proof hash

How Acre Works

Lender's Perspective: 4-Step Integration

flowchart LR
    subgraph CONFIG["1. Configure"]
        SELECT["Select proof modules<br/>(income, tenure, rating, identity)"]
        POLICY["Set risk thresholds<br/>(tier cutoffs, loan limits)"]
    end

    subgraph WORKER["2. Worker Verifies"]
        QR["QR scan to Uber/Swiggy"]
        PROOF["ZK proof generated locally<br/>(zero custody)"]
    end

    subgraph COMPUTE["3. Score & Store"]
        SCORE["Blue Score computed<br/>(explainable metrics)"]
        CHAIN["Proof hash + consent<br/>logged on Algorand"]
    end

    subgraph QUERY["4. Query & Disburse"]
        API["get_eligibility(wallet)<br/>→ tier, limit, timestamp"]
        DISBURSE["Approve or reject<br/>in real-time"]
    end

    CONFIG --> WORKER
    WORKER --> COMPUTE
    COMPUTE --> QUERY

    style CONFIG fill:#1a1040,stroke:#a78bfa,color:#f8fafc
    style WORKER fill:#0a1a2a,stroke:#00e5ff,color:#f8fafc
    style COMPUTE fill:#2a1a1a,stroke:#ec4899,color:#f8fafc
    style QUERY fill:#0a2018,stroke:#34d399,color:#f8fafc

Step-by-Step: Worker to Lender

sequenceDiagram
    participant L as Your NBFC
    participant F as Acre Platform
    participant W as Gig Worker
    participant R as Reclaim Protocol
    participant SC as Smart Contract
    participant I as Algorand Indexer

    L->>F: Configure policy (income, tenure, rating, activity, completion)
    L->>F: Set risk appetite and lender rules
    
    W->>F: Connect wallet
    F->>R: Create QR session
    W->>R: Scan QR, login to Uber
    R->>W: Generate ZK proof (locally)
    W->>F: Submit proof
    F->>F: Verify ECDSA signature
    F->>F: Extract metrics (income, tenure, rating, activity, reliability)
    F->>F: Compute Blue Score (300-900) + affordability limit
    F->>SC: Store proof hash + score + consent outcome
    SC->>I: Log on-chain
    
    L->>SC: Query get_eligibility(worker_wallet)
    SC-->>L: Blue tier, credit limit, proof timestamp
    L-->>W: Approve or price loan using own policy
    L->>I: Audit trail ready for RBI inspection

What You Configure

The Lender Console

No coding. Visual configuration. See impact before deployment.

┌─ SELECT PROOF MODULES ──────────┐
│ ✓ Income (₹20k–₹40k–₹60k)      │
│ ✓ Tenure (3mo–6mo–12mo)        │
│ ✓ Rating (4.0–4.5–4.8)         │
│ ✓ Completion Rate              │
│ ✗ Crypto Holdings              │
└────────────────────────────────┘

┌─ REVIEW SCORE DRIVERS ──────────┐
│ Income Stability ........ 30%    │
│ Consistency / Tenure .... 25%    │
│ Platform Rating ......... 20%    │
│ Activity Volume ......... 15%    │
│ Completion Reliability .. 10%    │
└────────────────────────────────┘

┌─ SET LOAN PRODUCTS ─────────────┐
│ Blue Prime (700+): best pricing │
│ Blue Plus (530-699): standard   │
│ Blue Basic (<530): entry tier   │
│ Limits bounded by income, DTI,  │
│ and product-level caps          │
└────────────────────────────────┘

Pre-Deployment Impact Preview

See how your thresholds affect approval rates before going live:

Your current policy:
└─ 1,000 gig-worker applicants
   ├─ 180 qualify Blue Prime (best pricing)
   ├─ 520 qualify Blue Plus (standard pricing)
   └─ 300 qualify Blue Basic or require review
   
   Portfolio impact: lender-specific credit policy
   Audit trail: proof hash + score + limit + timestamp

Architecture

3-Layer System (You Own Layer 3)

graph TB
    subgraph LAYER1["Layer 1: Proof Engine"]
        RECLAIM["Reclaim zk-TLS<br/>(income, tenure, rating)"]
        DID["DID / DigiLocker<br/>(identity)"]
        NOIR["Noir ZK Circuits"]
    end

    subgraph LAYER2["Layer 2: Feature & Score (Off-Chain)"]
        FEATURES["Feature Extraction"]
        SCORE["Blue Scorecard<br/>(explainable metrics)"]
        DASHBOARD["Worker Dashboard<br/>(score, tier, profile)"]
    end

    subgraph LAYER3["Layer 3: Decision & Compliance (You Control)"]
        CONSOLE["Your Lender Console<br/>(policy config)"]
        CONTRACT["Acre Smart Contract<br/>(Algorand App ID: 764223486)"]
        INDEXER["Algorand Indexer<br/>(audit trail)"]
    end

    RECLAIM --> FEATURES
    DID --> FEATURES
    NOIR --> FEATURES
    FEATURES --> SCORE
    SCORE --> DASHBOARD
    DASHBOARD --> CONSOLE
    CONSOLE --> CONTRACT
    CONTRACT --> INDEXER

    style LAYER1 fill:#1a1040,stroke:#a78bfa,color:#f8fafc
    style LAYER2 fill:#0a1a2a,stroke:#00e5ff,color:#f8fafc
    style LAYER3 fill:#0a2018,stroke:#34d399,color:#f8fafc

Layer	What It Does	Who Controls It
Proof Engine	Generates ZK proofs locally (zero custody)	Reclaim + Algo community
Feature & Score	Computes features and Blue Score	Acre platform
Decision & Compliance	Your policy, your audit trail	YOU (Lender)

Critical: You never see raw data. You only see: proof hash + score + eligibility outcome.

Smart Contracts

App ID (TestNet): 764223486

What the Contract Does

@application.internal()
def verify_income(
    user_wallet: str,
    tier: uint64,
    credit_limit: uint64,
    timestamp: uint64,
    proof_hash: bytes,
    rider_count: uint64,
    rider_rating: uint64,
    platform: str,
    score: uint16,
    income_bucket: uint8,
    tenure_bucket: uint8,
    completion_bucket: uint8,
    rating_bucket: uint8,
    source: str,
    plausibility_flags: uint8,
    monthly_earnings: uint64,
    tenure_months: uint64,
    completion_rate: uint64,
) -> None:
    """
    Store worker's verified score and credit tier.
    Called by Acre backend after proof verification.
    Only designated verifier can write.
    """
    worker_state = local_state(acct := TxnFields.sender())
    worker_state['proof_hash'] = proof_hash
    worker_state['score'] = score
    worker_state['tier'] = tier
    worker_state['credit_limit'] = credit_limit
    worker_state['monthly_earnings'] = monthly_earnings
    worker_state['tenure_months'] = tenure_months
    worker_state['completion_rate'] = completion_rate
    worker_state['timestamp'] = timestamp

@application.external(read_only=True)
def get_eligibility(address: str) -> uint64:
    """
    Query worker's current stored credit limit.
    Permissionless — any lender can call.
    """
    worker_state = local_state(address)
    return worker_state['credit_limit']

How You Use It

# 1. After Acre backend verifies a worker's proof
algosdk.send_atomic(group=[
    txn.verify_income(
        proof_hash="0x1a2b3c...",
        tier=2,
        credit_limit=25000,
        score=685,
        monthly_earnings=36000,
        tenure_months=18
    )
])

# 2. When an applicant submits a loan request
limit = contract.get_eligibility(worker_wallet)
profile = contract.get_full_profile(worker_wallet)
score = profile.score

if score >= 530 and limit > 0:
    print(f"Review ₹{limit} eligibility under your policy")
else:
    print("Decline or request updated verification")

Project Structure

This repository keeps the demo product, backend verifier, scoring model, and Algorand contract in one workspace.

.
├── server.js              # Main Express API: proof verification, DigiLocker mock flow, scoring, contract writes
├── scoring.js             # Blue Score model, synthetic signal generation, tier and credit-limit policy
├── rider-server.js        # Smaller legacy/mock rider API used for isolated demos
├── advisorPrompt.js       # System prompt for the Acre advisor/chat experience
├── acre-web/              # Primary React + Vite frontend for workers, lenders, dashboards, and simulator
│   ├── src/               # Pages, components, wallet context, API client, Reclaim integration
│   ├── public/            # Static browser assets
│   ├── demo/              # Frontend demo screenshots
│    
├── contracts/             # Algorand contract source, compiled TEAL, ABI, deploy/call scripts, TestNet app data
├── income-verifier/       # Standalone React proof-verifier experiment kept separate from the main Acre UI
├── docs/                  # Product, API, architecture, score, verification, GTM, and flow documentation
├── demo/                  # README/marketing screenshots for the end-to-end product walkthrough
├── digi-aloplonk.md       # DigiLocker + AlgoPlonk integration notes and implementation plan
├── technical.md           # Technical architecture and implementation reference
└── SCREENS.md             # Visual walkthrough of the Acre product surfaces

The normal local loop is: run server.js for the backend, run acre-web for the app, and use contracts/ only when rebuilding or redeploying the Algorand verification contract.

🛠️ Tech Stack

Area	Technologies
Framework	Configurable, privacy-preserving underwriting
Proof Engine	Reclaim Protocol (zk-TLS), Noir ZK Circuits, DID-ready
Frontend	React 18, TypeScript, Vite, Tailwind CSS
Backend	Node.js 20+, Express, Algorand SDK
Smart Contracts	Algorand Python contract, ARC-4 ABI
Database	Supabase (workflows, audit logs)
Blockchain	Algorand (sub-3s finality, ~₹0.02 per verification)

Why Algorand

Property	Why It Matters to You
⚡ Sub-3s Finality	Real-time loan decisions during customer session
💰 ~0.001 ALGO/tx	Verification costs ~₹0.02; doesn't eat margins on ₹5k loans
🏛️ Deterministic Execution	Credit rules execute identically every time—RBI compliance critical
🗂️ ARC-4 + Indexer	Clean SDK for your team; immutable audit trail for regulators
🌱 Carbon Negative	ESG alignment for impact-focused investors and regulators

Screenshots

See the full visual walkthrough of Acre's lender and worker flows: SCREENS.md

Quick preview:

Lender Console: Configure your scorecard, set thresholds, see portfolio impact
Worker Proof Flow: Identity → QR scan → ZK proof → Blue Score → Dashboard
What-If Simulator: See how actions unlock better credit tiers
Lender Dashboard: Monitor workers, track approvals, manage risk

Getting Started

For Lenders (2-Hour Onboarding)

Step	Time	Action
1. Demo	30 min	See Acre in action + ask compliance questions
2. Configure	45 min	Build your scorecard in Lender Console (no code)
3. Test	30 min	Run 10 test workers through end-to-end
4. Deploy	15 min	Flip switch to production (TestNet → MainNet)

Result: Live gig-worker underwriting in 2 hours. Your audit trail baked in.

For Workers (60 Seconds)

Visit acre-web-three.vercel.app
Connect Pera or Defly wallet
Scan Reclaim QR → Login to Uber/Swiggy
See your Blue Score + credit tier
Share proof with lenders (your choice)

Fund TestNet wallet: Algorand dispenser

Developer Setup

Prerequisites

Node.js 18+
npm 9+
Docker (optional, for LocalNet)

Clone and install

git clone https://github.com/somehowliving/acre.git
cd acre

cd acre-web && npm install
cd .. && npm install

Environment configuration

Frontend (acre-web):

cp .env.example .env.local

Variable	Required	Purpose
`VITE_RECLAIM_APP_ID`	Yes	Reclaim protocol
`VITE_RECLAIM_APP_SECRET`	Yes	Reclaim secret
`VITE_BACKEND_VERIFY_URL`	Yes	Backend endpoint
`VITE_ALGORAND_APP_ID`	Yes	`764223486` (TestNet)
`VITE_ALGOD_SERVER`	Yes	Algorand RPC

Backend (repo root):

cp .env.example .env

Variable	Required	Purpose
`APP_ID`	Yes	Acre contract app ID
`ALGOD_SERVER`	Yes	Algorand RPC
`VERIFIER_MNEMONIC`	Yes	Proof signer account

Run locally

Terminal 1 - Backend

npm start
# Listening on http://localhost:3001

Terminal 2 - Frontend

cd acre-web
npm run dev
# Listening on http://localhost:8080

Privacy & Compliance

DPDP Act 2023 Alignment

Every feature is designed to minimize your DPDP liability.

Principle	Your Protection
Data Minimization	Zero raw financial data on your servers; only proof hash + score
Purpose Limitation	Credit eligibility only; no cross-selling, no aggregation
Storage Limitation	Acre stores nothing; you store only tier/score/limit/timestamp
Consent-Based	Worker explicitly approves every proof; logged on-chain
Verifiability	Algorand Indexer provides immutable audit trail for inspections
Right to Erasure	On-chain state nullifiable; zero off-chain raw data

Your Audit Trail

✓ Algorand Indexer logs every proof verification (consent + score, no PII)
✓ Suitable for RBI Digital Lending inspection
✓ Suitable for DPDP Act compliance audit
✓ Export for your compliance team in 2 clicks

Business Model & GTM

Detailed pricing, unit economics, roadmap, and partnership strategy: GTM.md

References

Team

zkFarmers — Building regulatory-safe credit infrastructure for emerging markets

Member	Role	GitHub
Nidhi Prajapati	Blockchain & ZK Engineer	@somehowliving

8M+ gig workers. 0 custody. ₹250Cr DPDP penalty avoided.
Acre is the underwriting framework regulators want.

Built for India's future of finance

Name		Name	Last commit message	Last commit date
Latest commit History 30 Commits
.agents/skills		.agents/skills
.claude/skills		.claude/skills
.codex		.codex
acre-web @ d66c4b9		acre-web @ d66c4b9
contracts		contracts
demo		demo
docs		docs
income-verifier		income-verifier
.env.example		.env.example
.gitignore		.gitignore
.gitmodules		.gitmodules
.mcp.json		.mcp.json
AGENTS.md		AGENTS.md
CLAUDE.md		CLAUDE.md
LICENSE		LICENSE
README.md		README.md
SCREENS.md		SCREENS.md
advisorPrompt.js		advisorPrompt.js
digi-aloplonk.md		digi-aloplonk.md
package-lock.json		package-lock.json
package.json		package.json
rider-server.js		rider-server.js
scoring.js		scoring.js
server.js		server.js
technical.md		technical.md

Folders and files

Latest commit

History

Repository files navigation

Acre — Privacy-Preserving Underwriting Framework for Gig Workers

Table of contents

The Problem

RBI 2025 Blocked You. DPDP Threatens Your Balance Sheet.

The Cost of Silence

The Solution

Acre is not a credit score. It is a configurable, privacy-preserving underwriting framework sold to lenders.

What Changes for You

How Acre Works

Lender's Perspective: 4-Step Integration

Step-by-Step: Worker to Lender

What You Configure

The Lender Console

Pre-Deployment Impact Preview

Architecture

3-Layer System (You Own Layer 3)

Smart Contracts

What the Contract Does

How You Use It

Project Structure

🛠️ Tech Stack

Why Algorand

Screenshots

Getting Started

For Lenders (2-Hour Onboarding)

For Workers (60 Seconds)

Developer Setup

Prerequisites

Clone and install

Environment configuration

Run locally

Privacy & Compliance

DPDP Act 2023 Alignment

Your Audit Trail

Business Model & GTM

References

Team

About

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Packages