Bangongyi AI for Bank · Yuerong
SME acquisition:
see clearly, then lend.
Yuerong is Bangongyi AI’s SME customer-acquisition agent for commercial banks. Once a business authorizes access, it aggregates tax, invoice, registry, court and transaction data into an operating profile, checks it line by line against each product’s eligibility rules, and returns recommended products, segment tags and the reasons a product is not recommended. Approval and disbursement stay with the bank; the bank pays on results.
A technology company · Wuhan · 6 years · small enterprise
- Tax credit rating
- Grade A · general taxpayer
- Invoiced, last 12 months
- ¥18.6M · +12% YoY
- Zero-invoice months
- 0 · no void or red-letter anomalies
- Legal-rep change, 6 months
- None
- Court risk signals
- No defaults · no enforcement
Sample figures. Real reports are generated from tax, invoice, registry, court and transaction data the business has authorized.
Core capabilities
From authorized data to an explainable eligibility verdict.
The agent does not approve loans. It does the three most labor-intensive jobs before approval: see the business clearly, lay out the risk signals, and check every product rule.
Tax and invoices across every time window
After authorization, tax and invoice data are aggregated into three kinds of indicators: scale, trend and continuity.
- Tax credit rating, taxpayer type, violation count, current arrears and late fees
- Taxable sales, tax paid and invoiced amounts for the last 12 and 24 months
- Invoice trends for 3 / 6 / 12 / 13–24 months: YoY, MoM, valid count, zero-invoice months, void and red-letter invoices
- Month-by-month invoicing and three years of monthly tax payments, drillable from the summary
See the entity, and who it trades with
Entity stability, court signals and trading structure in one report, ready for the relationship manager to verify.
- Shareholder type, subscribed capital, ownership; registry and legal-representative changes
- Defaulter listings, enforcement, court notices and judgments with amounts, tagged with the company’s role in each case
- Three-year financial highlights, insured headcount, bank-tax interaction records
- Top upstream and downstream counterparties ranked by last-year transaction value
One company, one verdict per product
Company features are checked against each product’s eligibility rules; the output is richer than yes or no.
- Grouped by invoice loan, tax loan, merchant loan, logistics and leasing, each marked recommended or not now
- Dimensions include business location, invoice volume, applicant age and account status
- Segment tags (premium, supplier segment) and a preliminary credit limit
- Every “not now” carries a reason: region, scale or age outside the product’s rules
Workflow
Six steps from authorization to the bank’s hand-off.
Authorization
The business applies through Bangongyi, a channel partner or WeChat and authorizes use of its operating data. Nothing is processed without it.
Data aggregation
Tax and invoice, registry and court, financial, social-insurance and transaction data are gathered into three layers.
Feature analysis
Scale, trend and risk signals across time windows: sales, invoicing, tax, YoY and MoM, operating continuity.
Product matching
Company features are compared with each product’s rules and a status is returned per product.
Explained results
Status, segment tags, preliminary limit and reasons, in one report shared by the business and the bank.
Bank hand-off
The bank completes approval and disbursement in its own process; the service fee is settled on credit and disbursement results.
Use cases
At every stage of acquisition.
Pre-screening for inclusive finance
Before a relationship manager makes contact, surface the businesses that fall within the product’s region, scale and other rules, with the reasons attached.
Cross-product matching
One authorization and one profile, checked against several products at once; unsuitable products come with explicit reasons, cutting manual comparison.
Channel-driven referrals
Over 600 channels, WeChat and offline SaaS partners guide businesses with financing needs to authorize and apply, and the matched list is pushed to the bank.
Post-loan monitoring
Authorized data keeps updating; swings in operations or new court signals raise alerts, a second pair of eyes for post-loan supervision.
Co-built models
Build or tune matching models for the bank, co-maintain allow and block lists, and co-develop SME customer models so good borrowers get cheaper credit.
Value for banks
Pay on results. Explainable verdicts. Compliant data.
Pay on results
The bank pays on credit and actual disbursement outcomes, with no upfront software fee. Our return is tied to the business the bank actually books.
Focus on eligible segments
Recommendations come with eligibility reasons, so relationship managers pursue only businesses inside the product’s rules; the analysis and court details are one click away for review.
Every verdict is traceable
The same company gets different statuses under different products, each tied to a specific product rule and shown as a reason and a tag, not a black-box score.
Compliance foundation
Business authorization is the precondition; data is encrypted in transit and at rest; MLPS level 3 certified; our own Kaopu large model is registered with the CAC; licensed for value-added telecom services.
How we prove it works
Four measures, all reconciled against the bank’s own records. Platform scale never substitutes for technical effect.
- Eligibility agreement rate
- Share of cases where the system’s pre-screen matches the bank’s review under the same product rules and data date
- Effective conversion
- Recommended → valid application → credit approved → disbursed, de-duplicated at each stage
- Processing efficiency
- Handling time for comparable customers, manual process versus system-assisted
- Acquisition cost
- Unit cost per approved or disbursed customer within the same business scope
Partners
15 partner banks; four list us as a tier-1 service provider at head-office level.
Partnerships range from direct-signed lending to co-built matching models and segment list delivery. The banks below have established the relationship at head-office level.
The data behind the matching
Credentials
Bangongyi AI for Bank · Yuerong
Want to run a real match
on your own products?
Send us one product’s eligibility rules, or a batch of authorized businesses, and we will produce a match report from real data, reconciled on the four measures above.