Your Data. Your AI. Zero Setup.
We pre-train a lightweight AI assistant on your purchased dataset, deploy it to your dedicated server, and deliver plain-English analytics in seconds.
From $300/mo all-inclusive • 24-48 hour setup • Your data never leaves your server
✓ No external APIs
✓ Pre-trained on your schema
🎯 The Data Analytics Gap
You bought high-value B2B data. Now you need to:
- Segment audiences for targeted campaigns
- Filter by complex firmographic criteria
- Export exact CSV lists for outreach
But you face three barriers:
- ❌ No SQL expertise — Your team can’t write complex queries
- ❌ No data science resources — Hiring an analyst costs $8K+/mo
- ❌ Generic AI tools fail — ChatGPT doesn’t know your schema, can’t execute queries, and sends your data to external servers
The Result? Valuable datasets sit unused, or analysis takes days of manual work.
Our Solution: A custom AI assistant, pre-trained on your exact dataset + schema, running 100% on your server. Ask questions in plain English → get validated SQL → download exact CSV results in 10-50 seconds.
⚙️ How It Works
You Purchase + Configure
- Select your dataset (or upload your own)
- Choose cloud provider: Hetzner / OVH / AWS / Bring Your Own
- We handle schema extraction + pre-training workflow
We Pre-Train + Deploy (24-48 hours)
- Extract your database schema + generate few-shot examples
- Optimize Qwen-1.8B/7B or Llama-3-8B (4-bit quantized) for your data structure
- Deploy to your dedicated server: database + AI model + chat UI + monitoring
- Deliver admin credentials + user guide + 60-min onboarding call
You Ask, Get, Download
You type: “Show me marketing directors in Berlin”
AI generates: SELECT * FROM contacts WHERE job_title ILIKE ‘%marketing director%’ AND city = ‘Berlin’ LIMIT 10000;
Result: 1,247 records • [Download CSV]
• Total time: 10-50 seconds
• Zero data egress
🔧 Technical Specifications
AI Model & Performance
- Models: Qwen-1.8B/7B or Llama-3-8B (4-bit quantized)
- Inference Engine: vLLM for optimized throughput
- Response Time: 10-50 seconds for analytical queries on 10M+ row datasets
- Context Window: Schema-aware prompting with few-shot examples
Data Infrastructure
- Database: PostgreSQL 15+ or ClickHouse (your choice)
- Storage: NVMe SSD, sized to your dataset (10GB-200GB+)
- Backup: Automated snapshots with cross-region replication option
Security & Sovereignty
- Query Validation: Whitelist allowed tables/columns; block DROP/DELETE; enforce LIMIT clauses
- Audit Logging: Complete trail of all queries, users, timestamps
- Zero External Calls: No API calls to OpenAI, Anthropic, or other third parties
- Your Keys, Your Cloud: Infrastructure provisioned under your account; we never hold credentials
Monitoring & Support
- Health Checks: Automated alerts for model performance, disk space, query latency
- First 30 Days: Unlimited support included (UI tweaks, query optimization, training)
- Ongoing: Optional retainers for model upgrades, additional datasets, SLA guarantees
Note: This service requires Pathway B (Scalable Server) infrastructure. Learn about Data Workstations →
🎯 Use Cases: What You Can Ask
🎯 Marketing Segmentation
→ 842 records → [Download CSV]
ABM campaign targeting, event invitation lists
🎯 Sales Prospecting
→ 1,103 companies
Territory planning, partner outreach
🎯 Market Research
→ Summary stats + report
Strategy decks, client presentations
🎯 Account-Based Marketing
→ Precision segment
LinkedIn ads, direct mail, sales sequencing
🎯 Compliance & Audit
→ Compliance report
GDPR/CCPA documentation, internal audits
Note: The AI can answer questions about columns that exist in your dataset. It cannot invent data, predict outcomes, or query external sources.
💰 Pricing
🟢 Starter
$300/mo
- Up to 10M records
- RTX 4070-class GPU (12GB VRAM)
- PostgreSQL database
- Qwen-1.8B model
- Basic monitoring + alerting
First 30 days unlimited support included
🔵 Professional
$400/mo
- Up to 50M records
- RTX 4090-class GPU (24GB VRAM)
- PostgreSQL or ClickHouse
- Qwen-7B or Llama-3-8B model
- Advanced monitoring + query optimization
- Priority email support
🟣 Enterprise
$500/mo+
- 50M+ records (custom quote)
- Multi-GPU or A100 option
- Custom model fine-tuning on your data
- SSO integration + advanced audit logs
- SLA guarantees + dedicated support channel
- ✓ Cancel anytime
- ✓ Bring-your-own-cloud: 15% discount (you pay cloud provider directly)
- ✓ 24-48 hour setup included in all tiers
- ✓ First month unlimited support included
This service builds on our Scalable Server Workstations. If you prefer manual analysis without AI, explore Pathway B options.
❓ Frequently Asked Questions
Q: What if I don’t know what questions to ask?
We provide a library of example queries tailored to your dataset + a “query builder” helper with suggested filters. Our support team can also suggest analyses based on your business goals during onboarding.
Q: Can the AI answer any question about my data?
It can generate queries for columns that exist in your dataset. It cannot: invent data that isn’t present, predict future outcomes (requires separate ML modeling), query external sources or APIs, or answer ambiguous questions without clarification.
Q: What if the AI generates a slow or expensive query?
Our validation layer blocks full-table scans and enforces LIMIT clauses by default. You can also set custom query timeout thresholds. All queries are logged for review.
Q: How is this different from ChatGPT + my CSV?
Three key differences: 1) Schema awareness: Our AI is pre-trained on your exact database structure. 2) Execution: It validates and runs the query against your live data. 3) Sovereignty: Nothing leaves your server.
Q: Can I start without AI and add it later?
Yes. You can begin with a standard Pathway B Data Workstation and upgrade to AI assistant anytime. Pre-training takes 24-48 hours; no data migration required.
Q: What support do I get?
All tiers include first 30 days unlimited support (UI tweaks, query optimization, training). Professional+ tiers include priority email support. Enterprise includes dedicated channel + SLA guarantees.
Q: Is my data safe during pre-training?
Yes. Pre-training happens in a secure, isolated environment. We extract only schema metadata + anonymized sample rows. Your full dataset never leaves your control. We sign NDAs and delete all temporary artifacts post-deployment.
🔍 AI Assistant vs. Standard Data Workstation
| Feature | Standard Workstation (Pathway B) | + AI Data Assistant |
|---|---|---|
| Query Method | Manual SQL or pre-built filters | Natural language → AI-generated SQL |
| Technical Skill Required | Basic SQL helpful | None for basic queries |
| Setup Time | 24-48 hours | 24-48 hours + 2-4 hours pre-training |
| Best For | Teams with SQL expertise; complex custom workflows | Teams needing rapid segmentation; non-technical users |
| Pricing | $150-400/mo (infrastructure only) | +$350/mo for AI layer (all-inclusive) |
| Data Sovereignty | ✓ Your server, your control | ✓ Same + zero external API calls |
Both options use the same secure, sovereign infrastructure. The AI add-on is purely a usability layer on top of your existing workstation.
Compare all workstation options →
🚀 Ready to Turn Your Data Into Instant Insights?
- Share your dataset (NDA available) + describe your ideal workflow
- Receive proposal with architecture, timeline, and fixed investment
- We build remotely; you review via screen-share demos
- Handover + training + 30 days unlimited refinement support
Setup time: 24-48 hours • Cancel anytime • Your data never leaves your server
