— Our services
Three AI solutions,
carefully defined
Each service is designed around a specific problem type. We'll help you identify which one fits your situation — and what to expect from the engagement.
← Back to Home— How we work
Our methodology
Every Pulsemind engagement follows a consistent structure, regardless of which service you've chosen. We begin with discovery, move through design and development with regular check-ins, and close with a documented handover. There are no surprises and no phase that exists only to extend the contract.
01
Discovery
Data review and goal clarification
02
Design
Scoped proposal and approach
03
Delivery
Build, test, and calibrate
04
Handover
Training, docs, and knowledge transfer
— Service 01
Sentiment Analysis Platform
We build sentiment analysis tools that process text from surveys, reviews, social media, and internal channels to surface emotional patterns and trends. The platform provides visual dashboards and regular summary reports so your team can respond thoughtfully to shifting sentiment. Configuration is handled by us — your team simply logs in and explores.
What's included:
Process steps:
- 1. Data source inventory and access setup
- 2. Model selection and Cantonese/English tuning
- 3. Dashboard configuration and user testing
- 4. Team onboarding and documentation delivery
— Service 02
AI-Powered Knowledge Base
Transform your organisation's accumulated knowledge — manuals, FAQs, past correspondence, internal wikis — into a searchable, intelligent system that employees and customers can query in natural language. We structure your content, build the retrieval architecture, and fine-tune response quality. The result is a knowledge base that understands context and provides relevant answers, reducing burden on your support team while improving information access for everyone.
What's included:
Process steps:
- 1. Content inventory and quality review
- 2. Retrieval architecture design and build
- 3. Query testing and response calibration
- 4. Interface handover and team training
— Service 03
Forecasting Model Development
We build statistical and machine learning forecasting models for organisations that need to anticipate demand, resource needs, revenue trends, or operational loads. The engagement begins with a careful review of your historical data, followed by model selection, training, and validation. Results are presented in a straightforward format with confidence intervals — so your team can make informed decisions. The model is delivered with documentation and a sixty-day calibration period.
What's included:
Process steps:
- 1. Historical data audit and preparation
- 2. Model selection and initial training
- 3. Validation and confidence interval analysis
- 4. Delivery, documentation, and 60-day calibration
— Choose wisely
Solution comparison
Not sure which service fits best? This table outlines the key differences to help you orient.
| Feature | Sentiment Analysis | Knowledge Base | Forecasting |
|---|---|---|---|
| Primary input | Text from reviews, surveys, social | Docs, wikis, FAQs, manuals | Historical numerical data |
| Primary output | Dashboards and trend reports | Natural language Q&A interface | Forecast with confidence intervals |
| Typical duration | 4–6 weeks | 6–10 weeks | 3–4 months (incl. calibration) |
| Starting price | HK$6,500 | HK$11,300 | HK$9,700 |
| Calibration period | 60 days | ||
| Best for | Brand, HR, CX teams | Support, ops, knowledge teams | Finance, supply chain, planning |
— Across all services
Technical standards we apply to everything
Security & Privacy
NDA before data access. PDPO compliance. Minimal data retention. Secure transfer protocols on every project.
Performance Standards
Each model or system is benchmarked against defined criteria. Known limitations are documented and communicated.
Support Approach
One business day response commitment. Named senior contact. Written milestone summaries. Bi-weekly check-ins on active projects.
Documentation
Every engagement includes technical and plain-language documentation. No project is closed until your team can operate the system independently.
Open Tooling
We build with open-source-first frameworks. No proprietary platform lock-in. You own and can maintain what we build.
HK Language Support
All text-based systems are tested and tuned for Cantonese, Traditional Chinese, and English — including code-switching content common in HK contexts.
Not sure which solution fits?
Tell us a bit about your data and what you're trying to understand or improve — we'll give you a candid view on which approach makes sense.
Discuss Your Situation