AI agents that live on your servers.
Train them on your documents, fine-tune them on your domain, deploy them through any channel — all without a single byte leaving your infrastructure.
The mid-market problem
EU AI Act enforcement
2 August 2026
High-risk AI systems must satisfy Articles 9-17 — documented risk management, human oversight, audit trails, conformity assessment, EU registration. Fines up to €35M or 7% of global revenue.
GDPR + Schrems II
since 2020
Routing personal data through US-based AI providers creates ongoing compliance risk. EU companies must justify each transfer or face DPA scrutiny.
No safe option
for mid-market
Hyperscalers are non-compliant by default. Mistral Studio targets Fortune 500. Hugging Face requires an in-house AI team. Mid-market gets nothing fitting.
Built for sectors that can't compromise on data
Legal & Compliance
Contract review against firm templates. Document confidentiality preserved. GDPR-compliant by architecture, not retrofitted.
Financial Services
Customer onboarding, KYC support, regulatory reporting. Schrems II compliant. EU data residency by design.
Manufacturing
Technical documentation search, supplier price comparison, RFQ analysis. Industrial IP stays inside your perimeter.
Healthcare
Internal protocol search, patient-facing FAQ in regulated languages, regulatory documentation. HDH-compatible.
Three steps from install to production
Day 1
Install
Cradle Server installs as a single binary on Linux or macOS. Pulls open-weight models. No external dependencies, no telemetry, runs offline from minute one.
Days 2-7
Train
Upload corporate documents — PDF, DOCX, XLSX, MD. Optional LoRA fine-tuning on company terminology. Pick from 10+ agent templates: support, legal, sales, HR.
Days 8-14
Deploy
Connect agents to Microsoft Teams, Slack, internal portals, web widgets, REST API. Operators monitor draft answers; approve, edit or reject — with full audit log.
What's inside Cradle
Eleven integrated layers — the moat against fragmented competitors.
- 01Multi-channel — Teams, Slack, Web, REST API
- 02EU AI Act compliance — audit log + governance
- 03Risk classifier — L1 rules + L2 LLM
- 04Agent router by skills and domain
- 0510+ agent templates with curated prompts
- 06RAG — multi-format parsers (PDF, DOCX, XLSX)
- 07sqlite-vec — embedded vector store
- 08Web crawler with robots.txt compliance
- 09Operator inbox — approve/edit/reject workflow
- 10LoRA pipeline — domain fine-tuning
- 11Model runner — llama.cpp, open-weight models
Choose your tier
All tiers — on-premise, annual subscription, no per-token fees.
Cradle Pro
Departments up to 30 users
- Up to 3 AI agents
- Up to 50 GB document storage
- Single channel integration
- Standard support (business hours)
- Self-service onboarding
Cradle Business
Companies up to 250 users
- Up to 10 AI agents
- Up to 200 GB document storage
- Multi-channel + REST API
- Extended support (8/5, 4h SLA)
- Onboarding + LoRA training included
- Per-seat add-ons available
Cradle Enterprise
Companies 500+ users
- Unlimited agents and data
- Distributed deployment
- Premium support (24/7, 1h SLA)
- Dedicated CSM
- Industry-specific customization
- Custom legal terms
Try before you commit
Two-week paid pilot on your data — €2,500. Conversion-credited toward annual contract.
Frequently asked questions
Does Cradle ever phone home?
No. After installation it runs entirely offline. The only outbound traffic comes from optional model downloads from Hugging Face — and only when you click "download" on a specific model. You can air-gap the server completely.
What hardware do I need?
For Cradle Business with Mistral 7B-class agents: 32 GB RAM, 16-core CPU, 200 GB SSD. GPU is optional but recommended. Specific guidance for each model is shown in the Models page warning system.
Which models can I run?
Any GGUF-format model: Mistral, Llama 4, Qwen 3, Gemma, Phi, DeepSeek. We curate a recommended list from public benchmarks (OpenLLM Leaderboard, LMSYS Arena), but you can add any custom URL.
How do I integrate with our existing tools?
Cradle exposes a REST API and supports Telegram, Slack, Microsoft Teams, web widgets, and custom channels. New channels typically take 200 lines of TypeScript. The HTTP API is stable and documented.
Can we fine-tune on our own data?
Yes, for Cradle Business and Enterprise. The platform includes a LoRA training pipeline running on your hardware (Apple Silicon or NVIDIA GPU). No data leaves your infrastructure during training.
What about EU AI Act compliance?
Cradle ships with audit logging, human-in-the-loop workflows, and risk classification by default. We provide templates for the documentation Article 11 requires. Final compliance is your responsibility — but the platform makes it possible rather than impossible.
Open source?
Cradle Community (limited) is MIT-licensed. Cradle Pro / Business / Enterprise are commercial. The architecture is open — you can read every line of code that runs on your servers.
Where are you based?
Distributed senior team, working remotely. The platform deploys on the customer's own infrastructure — data never leaves your perimeter.
Start a pilot in two weeks.
We work with you on real data, on your infrastructure. €2,500, two weeks, credited if you continue.