Three solutions, each with a clear scope and a fixed price
Voxelith doesn't offer undefined consultancy. Each service is built around a defined output, agreed success criteria, and a handover your team can actually use.
Back to HomeHow every engagement is structured
Discovery
We spend time understanding your data, your domain vocabulary, and what you actually need to do with the output. This shapes every subsequent decision.
Scope Agreement
A written document defines what will be built, what success looks like, what's out of scope, and what the handover will include. Signed before work begins.
Build & Review
Development proceeds in milestones with written updates. You're involved at review points — not just at the end. Feedback is incorporated before final delivery.
Handover
A walkthrough session with your team, full documentation, and 30 days of follow-up questions at no additional cost. You leave knowing how to use what you've been given.
Knowledge Graph Construction
We build knowledge graphs that map the relationships between entities in your organisation's data — connecting people, products, processes, documents, and concepts into an interconnected network. This structured representation enables more sophisticated querying, discovery, and reasoning than traditional databases allow.
Applications include enterprise search, regulatory compliance mapping, research acceleration, and customer relationship intelligence. We handle ontology design, data ingestion, graph construction, and provide a visual exploration interface.
What's included
- Domain discovery and ontology design workshops
- Data ingestion pipeline from your existing sources
- Graph database construction and configuration
- Visual exploration interface for end users
- Query templates for common use cases
- Full documentation and handover session
Typical process timeline
Week 1–2: Discovery and ontology design
Domain interviews, data audit, relationship mapping
Week 3–6: Pipeline and graph construction
Ingestion build, entity resolution, relationship population
Week 7–9: Interface and review
Exploration interface, query templates, client review rounds
Week 10–12: Testing and handover
Acceptance testing, documentation, walkthrough session
Project fee
HK$13,400
Text Mining and Extraction
Extract structured information from large volumes of unstructured text — contracts, emails, research papers, news articles, or regulatory filings. We build custom extraction pipelines that identify entities, dates, amounts, relationships, clauses, and other relevant data points specific to your domain.
The extracted information feeds into your existing systems or a dedicated dashboard. We focus on precision and recall metrics that match your tolerance for errors, and include a human review workflow for high-stakes extractions.
What's included
- Extraction schema design and review
- Custom pipeline build for your document types
- Human review queue integration for edge cases
- Benchmark testing against held-out documents
- Dashboard or system integration as agreed
- Documentation and handover session
Typical process timeline
Week 1: Schema design and samples review
Document type audit, entity definition, metric agreement
Week 2–3: Pipeline construction
Extraction rules, entity resolution, review queue setup
Week 4–5: Testing and integration
Benchmark runs, client review, integration or dashboard
Project fee
HK$7,800
AI Readiness Training
Prepare your team for a future that increasingly involves working alongside AI systems. This training programme covers practical topics — how AI models work at a conceptual level, how to evaluate AI-generated outputs, how to provide effective feedback to improve model performance, and how to recognise limitations.
Delivered over three sessions of ninety minutes each, the programme uses hands-on exercises with real AI tools. We adapt the difficulty and examples to your team's existing technical background. Designed to build practical competence and thoughtful scepticism.
Session breakdown
Session 1: How AI models actually work
Conceptual overview of how language models process information. No code required. Focus on mental models that help you work with AI tools more effectively.
Session 2: Evaluating and improving AI outputs
Hands-on practice evaluating AI outputs critically. Techniques for identifying errors, inconsistencies, and overconfident claims. How to give feedback that improves results.
Session 3: Recognising limits and building workflow
Where AI reliably falls short. How to design workflows that keep humans in the loop where it matters. Practical exercise applying principles to a real scenario from your domain.
Programme fee
HK$3,900
3 × 90-minute sessions
Choosing the right solution
Not sure which service fits your situation? This matrix may help narrow it down.
| Feature | Knowledge Graph | Text Mining | AI Training |
|---|---|---|---|
| Structured output delivered | |||
| Works with existing documents | |||
| Suitable for complex entity relationships | |||
| Identifies specific data points in text | |||
| Builds team capability | |||
| Visual interface included | |||
| Human review workflow | |||
| Typical duration | 8–12 weeks | 3–5 weeks | 2 weeks (3 sessions) |
Best for Knowledge Graph when...
You need to navigate complex relationships across many data sources — not just retrieve documents, but understand how entities connect across your organisation.
Best for Text Mining when...
You have a large body of documents and need to extract specific structured data from them — dates, parties, clauses, figures — at a scale your team can't manage manually.
Best for AI Training when...
Your team is starting to use AI tools but lacks the conceptual foundation to evaluate outputs critically or integrate them responsibly into their workflow.
Professional standards across all solutions
Data Security
NDA before data sharing. Compliant with Hong Kong PDPO. On-premise deployment available for sensitive engagements.
Documentation Standard
All deliverables come with documentation that covers design rationale, not just usage. Written for the people who will maintain it.
Quality Metrics
Success is measured against criteria agreed before work starts. Acceptance testing is part of every delivery, not an optional extra.
Post-Delivery Support
30 days of follow-up questions answered at no additional cost. We want your team to actually use what we've built.
Not sure where to start?
Share a rough description of what you're trying to do — even if it's not fully formed. That's usually enough to have a useful first conversation.
Get in Touch