AI that works for your team. Privately. Securely. Yours.
Your most sensitive data never leaves your building. Period. pdm.ai gives your organization a complete private AI workspace — deployed on your own servers, controlled by your team.
Deploy on your own servers. Own the full stack from day one. No vendor access, no lock-in.
AI at work today
into public AI tools
is completely unmanaged
Banning AI doesn't work.
Controlling it does.
Your staff are already using AI — on personal accounts, outside your visibility. The answer is smarter infrastructure that gives them great tools while keeping your data where it belongs.
The reality today
Employees use AI tools regardless of policy. Contracts, client files, internal reports — all flowing into public cloud services, outside your control and audit trail.
The pdm.ai approach
A complete private AI workspace running on your own servers. Upload your documents, index your knowledge bases, and ask questions in plain language. Everything — documents, queries, answers — stays on your infrastructure.
The outcome
Your team stays productive. Your compliance officer has the audit trail. Your IT team owns the stack.
Your team gets AI. Your compliance team gets the audit trail.
pdm.ai is a private AI workspace running entirely on your own servers. Documents are indexed locally, queries are answered locally, and every interaction is logged. Your team works. Your compliance team sleeps.
Security that's built in,
not bolted on.
The entire stack runs on your own infrastructure. Documents are indexed locally, the AI model runs locally, and the audit trail is stored locally. This is architecture, not policy. It cannot be accidentally bypassed.
Your most sensitive data stays home
Anything that could identify your clients, patients, or employees is handled entirely on your own infrastructure. No exceptions, no configuration required.
Wiki-first knowledge retrieval
The system synthesizes a structured wiki from your uploaded documents, then uses that wiki to answer questions with citations. Answers are grounded in your actual content — not hallucinated guesses.
Every decision is logged
A complete, tamper-proof audit trail: what was asked, which system handled it, and what was detected. When your auditor asks, the answer is already prepared.
Sensitivity controls built in
Document-level sensitivity labels and access controls are built into the platform. Restrict who sees what at the document level, with JWT auth and Argon2id passwords enforced throughout.
Your deployment, your control
You own the full deployment from day one. No vendor has access to your environment, your data, or your queries. You can migrate, fork, or extend it as you see fit. No lock-in, ever.
Runs entirely on your infrastructure
Deployed on servers you own and control. IT manages everything. No vendor has access to your environment, your data, or your queries. Ever.
Document formats supported
Upload and index documents in any common format — extraction and indexation happen automatically on your hardware.
- PDF, DOCX, PPTX, XLSX documents
- EPUB and plain text files
- Internal wikis and knowledge bases
- Answers include citations back to the source document
- Everything indexed and searchable on your own servers
Knowledge graph + concept map
The platform builds a structured knowledge graph from your documents — not just a flat keyword index.
- Interactive concept map: explore how topics connect across documents
- Cross-document linking: answers draw from the full knowledge base
- Wiki synthesis: a living wiki is maintained from your indexed documents
- Full audit trail: every query and every access decision is logged
A workspace your team
will actually use.
If the tool creates friction, people route around it. pdm.ai works like the AI tools your team already knows — just with the right data protection built in by default.
Search your own knowledge, privately
Index your documents, internal wikis, and knowledge bases. Ask in plain language. Get answers grounded in your actual content — on your own hardware, never in a cloud database.
Familiar from the first minute
Chat interface, document upload, saved history. No training required. If your team has used any modern AI tool, they already know how this works.
Compliance without extra work
Every query is logged, every sensitive routing decision flagged. Your compliance team gets a dashboard. Your auditor gets a report. Nobody changes how they work.
Deployed on your terms
Runs on your own servers. No cloud dependency, no vendor access, no lock-in. IT owns everything from day one. You can migrate, fork, or extend it as you see fit.
Your knowledge base. Your AI. Your servers.
pdm.ai gives your team an AI that knows your documents, answers with citations, and never sends anything outside your building. Everything runs on infrastructure you own and control from day one.
Private AI that fits
your organization.
The core platform is fully built and ready to deploy. Enterprise plans add specialized capabilities scoped to your specific needs.
What's live and ready today
- Private document search over your contracts, files, and knowledge bases — no data in the cloud
- Internal wiki indexation for intranets, SharePoint, and knowledge systems
- Private AI workspace with your branding and your login system
- Full audit trail and compliance dashboard
Enterprise capabilities
For organizations that need more than the standard platform. Some items are roadmap — scoped per engagement:
- Fine-tuning on your internal data, terminology, and processes
- Sensitive data interceptor for existing tools: Microsoft Copilot, ChatGPT, Claude
- Custom detection rules tailored to your industry and regulations
- Integration with SharePoint, DMS, and existing enterprise systems
- Architecture review, dedicated support, and custom SLA
Deployed today. Extended for your enterprise.
The private workspace — document search, wiki indexation, knowledge graph, and local AI — is fully built and ready to deploy on your servers. Enterprise plans add roadmap capabilities like cloud AI routing and fine-tuning, scoped per engagement.
Deployed and running.
Pilots open now.
Fully built and ready to deploy. Real organizations are already in conversation — with three pilot spots remaining at reduced pricing.
Where we stand today
Join as a pilot partner and get the first 3 months at reduced pricing. One tier per organization:
Tier 1 · Laptop
+ optional €60/mo · 3 months
Tier 2 · Team GPU
+ optional €150/mo · 3 months
One tier per organization. Pilot pricing applies to the first 3 months only.
You own the deployment.
Nothing is locked to us.
One-time setup covers deployment, configuration, and handover. Monthly support is optional — the deployment is yours to run independently from day one.
Any laptop · No GPU required
- 1–2 concurrent users
- Runs on any modern laptop — no server needed
- Private document search and wiki indexation
- Full local AI model included
- Deployment, configuration, and handover
- Email support included
1× GPU server · Up to 10 users
- Up to 10 concurrent users
- 1× GPU server (you provide or we advise)
- Full model quality — comparable to cloud performance
- Private document search, wiki indexation, knowledge graph
- Deployment, configuration, and team onboarding
- Priority support included
Multi-GPU · Unlimited users
- Unlimited users, scales with your hardware
- Fine-tuning on your internal data and terminology
- Sensitive data interceptor for Microsoft Copilot, ChatGPT, Claude
- Custom detection rules and enterprise integrations
- Architecture review included
- Dedicated support with SLA
Every tier includes the full platform
Private document search, wiki indexation, and a local AI model running on your own servers. Your model, your data, never leaving your building. Monthly support is optional — the deployment is yours from day one and you can run it independently forever.
Roadmap: intelligent routing
for your existing AI tools.
pdm.ai today is a complete private AI workspace. Next: an intelligent routing layer that lets your team keep using Copilot, ChatGPT, or Claude — while keeping sensitive data off those platforms automatically.
Shipped today ✓
- Private document search and wiki indexation
- Knowledge graph and interactive concept map
- Local AI model running on your own hardware
- Multi-user with JWT auth and sensitivity ACLs
- Full audit trail, backup/restore, license signing
- Vertical editions: Municipality, University, Law Firm, Healthcare, Personal Study
In development
- Intelligent routing layer: sensitive queries stay local, everyday queries go to the cloud AI your team already uses
- Interceptor for Microsoft Copilot, ChatGPT, and Claude
- Custom detection rule engine beyond defaults
- Fine-tuning on customer data and terminology
Already using Copilot or ChatGPT?
If your organization is committed to existing cloud AI tools and you want those tools to respect your data boundaries, get in touch. We're building the routing layer now and having early conversations about what it needs to do. If that's your situation, join the list below.
Inspect everything.
No black boxes.
Core tools are public. Your security team can review every line before anything touches your network. We believe security software that can't be inspected isn't security software.
Verifiable, not just "trusted"
We publish our tools on GitHub. Fork them, audit them, adapt them. The code that runs on your servers is the same code you can read.
↗ github.com/pdm-localStraight answers
to direct questions.
Which cloud AI providers do you support?
The shipped workspace runs entirely on your own infrastructure using a local AI model — no cloud API required. Cloud AI integration (OpenAI, Anthropic, Azure) is on the roadmap as part of the intelligent routing layer. If that's important for your team, see the roadmap section.
What data stays on my servers?
Anything that could be considered sensitive: personal identifiers, financial data, legal content, healthcare information, or whatever you define as sensitive for your organization. You set the rules. We enforce them automatically.
What happens when sensitive data is detected?
Everything is handled locally by default — because the entire workspace runs on your own servers. There's no outbound routing to worry about. The intelligent routing layer (which would evaluate queries heading to external cloud AI tools) is on the roadmap for organizations that also want to keep using Copilot or ChatGPT safely.
How does the local AI compare to cloud models?
For document search and knowledge retrieval from your own content, a well-configured local model is excellent. For open-ended general tasks, frontier cloud models are stronger. That tradeoff is one reason the intelligent routing layer is on our roadmap — so organizations can use both appropriately without compromising data boundaries.
How does this help with GDPR compliance?
All data is processed on your own infrastructure — no third-party processors involved, no cross-border transfers. The audit trail logs every query and every access decision, which satisfies most audit requirements we've encountered.
What if we need something custom?
Tell us what you need and we'll tell you honestly what's possible and what it takes. We've built custom AI deployments for public sector organizations, universities, and enterprises. Most non-standard requirements are solvable.
Built by practitioners,
not just theorists.
Everyone behind pdm.ai has deployed AI in real organizations: government, enterprise, and academia. This is what that experience looks like in a product.
Let's talk about
your situation.
Curious how this works for your organization? Tell us what you're trying to solve and we'll come back to you personally.
- ✓Personal reply within 3–5 working days
- ✓No data shared externally
- ✓Happy to do a short intro call first
- ✓Pilot spots still available at reduced pricing