Language

 
0
 
 

Local LLM / Private AI Services

Local LLM / Private AI Services — your data never leaves your company

Local LLM / Private AI Services

“Your data never leaves your company — and you still get AI.”

We deploy open-source large language models (LLMs) on your own server or in your own machine room. Customer lists, pricing, contracts, medical records, drawings — nothing ever goes to a third-party cloud.

Models: Qwen / Llama / DeepSeek and other open-source models. No vendor lock-in, no per-token API fees.

Data stays 100% on your infrastructure Open-source models · no API fees Connects to ERP / CRM / store / accounting On-site or remote delivery

Service levels (three starting points)

L1 · Local LLM deployment

HK$15,000 – 25,000
3 – 5 working days
  • Hardware assessment (is your machine enough?)
  • Model selection (open-source, matched to use case)
  • Installation, configuration and tuning (Ollama / vLLM)
  • Internal web chat interface
  • Measured speed and accuracy tests
  • 1-hour training session

L2 · + Private knowledge base (RAG)

HK$30,000 – 50,000
7 – 12 working days
  • Everything in L1
  • Ingest your documents (PDF / Word / Excel / web pages)
  • On-premise vector database (embeddings never leave)
  • Retrieval tuning (chunking, re-ranking)
  • Access control (who may ask what)
  • Accuracy test report
Popular use cases: support knowledge base, internal SOP Q&A, product manual lookup, contract clause lookup

L3 · + Workflow automation

HK$60,000 – 120,000
3 – 6 weeks
  • Everything in L2
  • Integration with your existing systems (API / database)
  • Auto-classify, auto-reply, auto-summarise, auto-posting
  • Security hardening: access control, audit logs, prompt-injection defence
  • Usage and performance dashboard
  • Architecture document and operations manual
Monthly care plan (optional, any level) HK$3,000 – 8,000 / month
System and hardware monitoring · evaluation and upgrade of new open-source models · ongoing knowledge-base ingestion · technical support · monthly usage report

Available standalone or bundled with AI

☁️ Nextcloud private cloud

HK$12,000 – 28,000
3 – 6 working days · 1,500 – 3,500 / month
  • Replaces Google Drive / Dropbox / OneDrive
  • File sync + web office (docs / sheets / slides)
  • Calendar, contacts, tasks, internal chat
  • Automatic backup + version history + recycle bin
  • 2FA, encryption, access logs
  • Wired into private AI: files feed the knowledge base directly

📊 Matomo private analytics

HK$8,000 – 18,000
2 – 4 working days · 1,000 – 2,000 / month
  • Replaces Google Analytics
  • Visitors, sources, pages, paths, funnels
  • Order / conversion / revenue tracking
  • 100% your own data, no third-party tracking
  • Friendly to GDPR and HK privacy law
  • AI reads the data and writes a weekly report

🎁 All-in-one bundle (best value)

HK$45,000 – 95,000
20 – 25% cheaper than buying separately
  • Private AI + private cloud + private analytics
  • Single sign-on
  • One backup strategy
  • One training session for all
  • All three systems connected

Why go private (most-asked questions)

Four reasons

  • Compliance — finance, healthcare, government: data may not leave the country
  • Privacy — customer lists, quotes and contracts must not go to a third-party API
  • Cost — at high volume, API fees can exceed building your own
  • Control — immune to vendor price rises and shutdowns

Compared with hiring

  • Hiring an AI specialist: HK$40–55K a month, HK$480–660K a year, 2–3 months to recruit
  • With us: L2 from HK$30–50K, delivered in 7–12 working days
  • The asset stays with you: people leave, the system and knowledge base remain
  • To scale: add a level, no new hires
First-year cost gap: ~HK$600K vs ~HK$80K

Next steps (three steps)

1 · Free assessment (30 minutes)

  • We look at your existing hardware and use case
  • You get a feasible plan and our view on the spot

2 · POC (1 week)

  • Small-scale test with your real documents
  • Decide only after you see results

3 · Quote and delivery

  • Priced by level and scale
  • Hardware: bring your own or we procure
Notes
1. Prices are indicative ranges; final pricing depends on scale, complexity and schedule
2. Hardware can be supplied by you or procured by us (at cost + 15%)
3. All open-source models, no vendor lock-in
4. Deployed on your own server or a nominated cloud; quotes exclude hardware, network and power