← Sticker Wall AGENT · DETAIL

AI-Powered Historical Bookkeeping and Invoice Cleaning and Archiving Outsourcing for Accounting Agencies: 200 RMB per Client, Monthly Earnings of 18,000 RMB

Workflow: Accounting agencies upload pending historical account sets, scattered invoice scans, and bank statement export files to

AGENT

Key Fields

FIELD STAMPS

IndustryFintech

RegionChina

ScaleSME

ChannelOnline

🔧 Workflow

Accounting agencies upload pending historical account sets, scattered invoice scans, and bank statement export files to a designated cloud folder every day as inputs. The system automatically calls OCR to extract invoice elements, automatically matches historical vouchers in the ERP, and flags missing, duplicate, or non-compliant anomalous invoices. Human intervention is only required to review anomalies, and upon confirmation, compliant electronic work papers, adjustment vouchers, and classification index tables meeting the 2026 tax archiving standards are output, which can be directly imported into accounting software to complete archiving.

🛠 Setup Requirements

First, it is necessary to analyze the data structures of ERP systems commonly used by different accounting agencies (such as Kingdee Cloud Star, Chanjet Good Accounting, Youyou Good Accounting, etc.) and design adaptable invoice classification and matching rules. Proficiency in low-code automation tools to call official open APIs is required, along with basic Python capabilities to handle dirty data, format conversions, and other exception scenarios. Initial setup takes about 10-15 working days, requiring full-link testing with 1-2 client ERPs first to verify recognition accuracy and archiving compliance before accepting batch orders. Data confidentiality agreements must be signed with clients in advance to mitigate historical accounting risks.

🧰 Toolchain

🔧 LaiShi Agent
🔧 Kingdee Cloud Star Open Platform
🔧 Tongyi Qianwen API
🔧 Tencent Docs

💰 Revenue

① Historical bookkeeping and invoice cleaning for accounting agencies billed per client (Main revenue): Accounting agencies pay a cleaning service fee per client, calculated as 200-300 RMB per client × monthly order volume of 90-100 clients = 18,000-30,000 RMB/month, consistent with the card-level monthly income of 18,000-30,000 RMB, accounting for approximately 100% of monthly income (derived by multiplying unit price by volume, based on self-reported merchant data without third-party verification); ② Review of anomalous invoices and preliminary tax compliance audits: Clients pay a manual review service fee per project, with quotes not publicly disclosed, and the number of completed orders is also undisclosed, leaving its share of total revenue unclear; ③ Batch account set monthly subscription: Agencies with hundreds of sets of historical accounts per month are offered packaged monthly subscriptions, though the price per set is unspecified, the number of contracted agencies cannot be verified, and its market share is not explained; ④ Opportunity item: Pipeline SaaS licensing: Referring to the business-finance platform's voucher automation rate reaching 90% (disclosed by the enterprise officially), licensing the invoice recognition-to-archiving pipeline annually to accounting agencies yields an unverified amount of revenue.

💸 Cost

Average monthly fixed costs are approximately 1,500 RMB, including about 600 RMB for cloud server rentals, 400 RMB for OCR API call fees, and 500 RMB for financial software open API call fees. There are no other hidden expenses, and initial trial-and-error costs account for only about 8% of monthly revenue.

⏱ Time Investment

Only 2 hours per day are needed to review anomalous invoices and coordinate client requirements; the rest of the workflow runs fully automatically with no on-site presence required. 1 hour per week is reserved for updating the tax rule base and ERP interface adaptation rules.

🚀 Getting Started

As a first step, beginners can apply for developer test permissions for Kingdee Cloud Star and Chanjet Good Accounting to run through the full process of invoice recognition to archiving for a single set of accounts. As a second step, they can contact 3-5 local accounting agencies to offer a free trial service for cleaning one set of historical books, leveraging result case studies to negotiate long-term cooperation. There is no need to build a proprietary system initially; third-party automation tools combined with manual review can be used first to lower trial-and-error costs.

🔑 Keys to Success

✅ Invoice recognition accuracy stabilized above 95%, with anomalous voucher recall rates exceeding 90%
✅ Familiarity with latest policy requirements such as the 2026 new Value-Added Tax Law and electronic invoice archiving specifications
✅ Ability to quickly adapt to data formats of ERP systems used by different accounting agencies
✅ Possession of basic financial and tax compliance risk preliminary audit capabilities, identifying historical non-compliance items in advance

⚠️ 风险

⚠️ Historical books involve risks of past non-compliance, requiring clearly defined liability boundaries in contracts to avoid bearing responsibility for clients' historical tax issues
⚠️ Low-clarity invoice scans may cause recognition errors leading to client claims, requiring prior agreement on error compensation rules
⚠️ Variations in regional tax archiving specifications may lead to rework, requiring rule adaptation tailored to different regions
⚠️ Risk of client historical financial data leakage, requiring confidentiality agreements and encrypted processing of stored data

📌 Real Cases

📌 Public reports by LaiShi Intelligent show that its cross-system data collection tool helps accounting agencies shorten historical bookkeeping processing hours by over 70%, with a certain outsourcing team undertaking batch cleaning of over 100 sets of historical books in a single month
📌 Chanjet's 2026 accounting industry survey shows that agencies using AI preprocessing tools increased per-capita historical bookkeeping processing volume by 3 times, while per-client processing costs decreased by 35%
📌 36Kr's "2026 Report on the Intelligence of Business, Finance, and Tax for Small and Micro Enterprises in China" points out that nearly 60% of accounting agencies plan to introduce AI tools in 2026 to handle historical accounting backlogs, with a market demand scale exceeding 2 billion RMB

Sources