Use case · Accounting software
Document AI, bank feeds and a universal connector for a Salesforce accounting product
Invoice and bank-statement extraction, automated bank feeds and a database-driven integration engine that connect a Salesforce accounting app to the outside world.
- Intelligent Document Processing
- Salesforce Integration
- Accounting System Integration
- AI Workflow Automation
Client details are kept confidential. This use case describes the problem and what we built.
The brief
A software company sells a finance and accounting application that runs natively on Salesforce. Its customers wanted documents read automatically, bank data fed in, and the app connected to e-invoicing networks, other ledgers and planning tools, without a long implementation project each time.
We built the document AI, bank-feed and integration services around the product, along with the Salesforce components that bring them into the app. We then consolidated the point-to-point integrations into one configuration-driven connector.
The challenge
What needed solving
Manual data entry from documents
Users of an accounting application built on Salesforce were keying in supplier invoices, remittance advice and bank statements by hand.
Bank and ledger data in other systems
Bank transactions, e-invoicing networks, other ledgers and planning data all had to flow into and out of customers' Salesforce orgs reliably.
One integration per customer did not scale
Writing a dedicated service for every new system and every new customer was slowing onboarding and adding maintenance cost.
The solution
What we built
Document AI service
A Spring Boot service uses Google Cloud Document AI to extract structured data from invoices, remittances and bank statements, with supplier-exclusion rules and an onboarding API. Results are pushed into the customer's Salesforce org.
Bank feed service
Scheduled jobs fetch bank transactions and post them to the target org, with feed monitoring for each customer.
Universal, database-driven connector
A generic engine connects any source to any target through a small connector interface (source, target and event source). Field mapping, child collections and aggregations are all set in configuration, so a new system is a thin plugin plus configuration rows.
Salesforce-side components
Lightning Web Components, Apex classes, triggers and custom objects in a Salesforce DX project embed the document and integration features in the accounting app.
Workflow automation and an AI assistant
n8n workflows handle email-driven document intake. A browser-extension assistant uses OAuth with PKCE to create records in business systems from web pages, emails and pasted text.
Containerised operations
Services and the document review UI run as containers on Kubernetes, with Flyway-managed schemas and an observability stack.
The outcome
What changed
Less manual keying
Invoices, remittances and statements are captured as structured data instead of being retyped.
Onboarding by configuration
New integrations and customers are added through connector plugins and configuration rows, not new services.
Integrations inside the user's workflow
Extracted data and bank feeds appear directly in Salesforce, where finance teams already work.
Technology used
- Java
- Spring Boot
- Google Cloud Document AI
- Google Cloud Pub/Sub
- PostgreSQL
- Flyway
- Quartz
- Salesforce Apex
- Lightning Web Components
- React
- TypeScript
- n8n
- Docker
- Kubernetes
Related use cases
Mapping hospital service catalogues to clinical codes, with reviewers in the loop
A mapping engine, hospital onboarding portal and LLM contract-rules extraction that turn messy hospital price lists into auditable, coded data for an insurer.
Read the use caseDeveloper onboarding where every training assignment gets an AI code review
A training and onboarding platform that links trainee pull requests to topics, sends them for AI review, scores them and keeps mentors in control.
Read the use caseAI timetable extraction from PDFs, images and Word files for a teacher-planning app
A teacher-planning platform that turns uploaded timetables and curriculum documents into structured, editable weekly plans using OCR and LLMs.
Read the use caseHave a similar problem to solve?
Share a few lines about your project. A senior engineer — not a salesperson — replies within one business day.