Document extraction and structured fields
Extract document content into agreed fields with source references. Missing or uncertain information goes through human review before use.
Software development for Hong Kong SMEs
Connect AI to a clear, testable business task.
Want AI to process documents, search company information or organise enquiries? ProApp starts with real work for Hong Kong SMEs: define inputs, expected results and error handling, then choose suitable models or conventional rules and build a system staff can operate and review.
What can we help you build?
Company knowledge search and assistants
Provide search, summaries and sources from authorised documents, with role-specific access and defined responses when no answer is available.
Enquiry classification and workflow integration
Organise messages, classify requests or draft responses, then connect to CRM or task lists. External messages and important updates follow agreed review steps.
Multi-step assistants and AI agents
Let an assistant search information, prepare drafts and call authorised tools in defined steps. Specify allowed actions, required approvals and when to stop after a failure.
Agree the scope before development begins.
The following work can be discussed. Included features, deliverables, fees and scheduling are agreed for your project.
- Choose a use case and representative samples, with expected results and test scenarios.
- Confirm data sources, access permissions, third-party model data handling and costs.
- Build processing flows, source records, human review and agreed business-tool integrations.
- Implement error handling and usage records, and document maintenance and the effects of model or data changes.
Estimate fees and scheduling around this scope.
What affects the quotation?
- Document/question types, sample quality, data volume and required outputs.
- Sources and role permissions, quality evaluation, human review and tool-action rules.
- Existing-system integration, deployment environment, expected request volume and processing time.
Assessment, development and handover
An initial discussion of requirements is free. Detailed assessments have an agreed scope and fee; development or repair specifies included features, tests, deployment and handover. Hourly work, fixed quotations and ongoing support follow the current pricing information and your agreed proposal.
Ongoing and third-party costs
Model/API usage, document and search storage, cloud computing and third-party costs are estimated for selected services and expected use. Monitoring, evaluations, data updates and maintenance have agreed responsibilities and fees. Prototype testing costs are distinct from production operation costs.
See pricing and engagement optionsAgree sample validation and production deployment separately.
Start with agreed samples or a proof of concept to check outputs, errors and review work. If suitable, scope production data connections, user access, monitoring and deployment separately. Passing samples does not establish the same performance for every document or question.
Confirm data-handling arrangements before use.
Identify which data each provider processes, storage locations, retention periods and whether data may be used for model training or service improvement. Confirm client requirements and selected service terms before choosing redaction, restricted access or other deployment arrangements; model policies differ.
How will completion be checked?
Use agreed samples and expected answers to check fields, sources, permissions and review flows. Test missing documents, unavailable answers and external-service failures. Agree acceptable results, manual exceptions and usage limits before production use.
What should you share in your first enquiry?
- One task you want AI to assist with and how you handle it today.
- Document or question-and-answer samples with sensitive data removed, plus correct results.
- Who will review results and which system should receive them.
Questions about this service
Can the AI system work without human checks?
It depends on the task. Sample tests define what can be automated and what needs review. Quotations, commitments and important data changes need an agreed confirmation process.
Can you build an AI chatbot connected to company documents?
We can assess retrieval from authorised information followed by source-based answers, often called RAG. Scope document sources, updates, user permissions and fallback messages or human follow-up when no answer is available.
Can I enquire before I have a specification or budget?
Yes. Describe the users, the problem and your target date. ProApp will outline a workable scope before confirming an estimate and scheduling. If a detailed assessment is needed, its work and fee are explained first.
How are handover and support arranged?
The proposal specifies source code, deployment, account ownership, operating instructions and acceptance checks. Ongoing maintenance, third-party subscriptions or usage fees and support hours are agreed for the project.
Related development guides
ProApp
Tell us what your business needs.
ProApp develops apps, websites, business systems and AI systems, and takes over AI-generated code for Hong Kong SMEs. New projects can start with requirements; existing systems start with a review of their current state. The founder directly handles technical decisions, implementation and communication.
Discuss this requirement