Your knowledge. Your environment.
Your team needs answers grounded in company knowledge, with clear boundaries around who can access what. That calls for an environment designed for your business, not just individual chat accounts.
Why it matters
Bring your team’s AI tools and knowledge together under your control.
01Employees use private ChatGPT/Claude accounts and paste customer data, contracts and internal documents into tools the company does not control. Nobody knows what left the building.
One environment with company login, roles and permissions, running on your own server. Data stays in your environment; you decide which models and sources are used.
Secure operations
A managed environment gives your team access to AI with controls suited to your workflows.
Common risk
Professionally managed
Staff paste confidential documents into public AI tools without clear oversight.
Your private environment uses self-hosted or EU-hosted models for confidential documents.
Shared API keys and logins make access difficult to control or revoke.
Each person has an individual account with MFA or SSO and documented role-based permissions.
Agents have unrestricted access to systems and can make changes without review.
We limit permissions to approved tasks and require human approval for sensitive actions.
Without logs or maintenance, agent activity and system health are difficult to assess.
We maintain logs, apply updates and monitor your managed environment.
On request, access is limited to approved devices on a private network, with no access from the internet.
We agree on scope and hosting for each project; ongoing care and consultation are available on request.
02Each person builds their own prompts, skills and workflows. They are duplicated, unoptimized, and disappear when someone leaves.
Proven prompts, skills and workflows are maintained centrally and made available to your team.
Data protection
Your confidential documents stay in a private environment with self-hosted or EU-hosted models.
Your data is hosted on servers in Germany or the EU.
We provide a data processing agreement (DPA).
Self-hosted open-source tools keep your data on your server without transfers to US providers.
Your data is encrypted in transit and at rest; we make regular backups and test recovery.
Individual accounts use MFA or SSO; role-based permissions are documented.
Export or delete your data at any time, without vendor lock-in.
Run the environment on a dedicated server in your preferred region or on-premises; on-premises requirements are discussed individually.
Retention and deletion rules and logged support access make data handling clear.
The measures are tailored to each project and its requirements.
03Business knowledge is spread across drives, email threads and individual people. New colleagues ask the same questions; answers depend on who you ask.
Handbooks, processes and documents are connected once; agents answer from the same source, with references and respecting access rights.
Self-hosted in your region or on-premises
Built on established open-source software
Your data stays exportable — no lock-in
One direct contact: David, from planning to operations
Proven building blocks
We set up and operate established tools instead of a proprietary black box. Your environment stays under your control, without lock-in.
Shared skills and workflows
An open-source AI agent for working with code, documents and tools. It powers shared skills and workflows.
Self-improving agent
An agent by Nous Research that creates and refines skills through use.
A workspace for your agents
An open-source interface for coding agents such as OpenCode, Codex and Claude Code, available on desktop, web and mobile.
Built for your company
Each agent has its own isolated container; the model and knowledge stay local to your dedicated environment.
Bring in company documents and define access by role and permission, so the agent works within agreed boundaries.
Document storage, CRM or ERP, and internal APIs are scoped and approved one by one. Actions that write or change data require separate approval.
Documented data flows make the setup understandable. People review answers before relying on them or using them to make decisions.
Model choice
Choose leading cloud models via business APIs or run open models on your own server. Switch providers without rebuilding your setup.
GPT models
Claude
Gemini
GLM
Kimi
DeepSeek
Mistral & open models
Model gateway
Use cases
Find answers in internal handbooks and policies. A person checks the answer when it matters.
Prepare drafts for replies and offers; your team reviews them before sending.
Summarize and compare documents or contracts, with a person reviewing important details.
Help new colleagues find their way through internal guidance. A colleague reviews important answers.
Gather context and prepare support tickets for a person to review before they are handled.
Research across approved internal knowledge sources. Your team checks the findings and sources when they matter.
Integrations
Connect approved document storage and intranet sources to make relevant information available.
Include email and calendar information where it supports the agreed workflow.
Connect CRM or ERP systems using only the records agreed together.
After review, selected database information can be made available to an agent.
Consider connections to internal applications as part of your individual workflow.
Connect APIs for an agreed purpose; actions that write or change data require separate approval.
What people ask
The things clients usually ask before we start working together - and how I think about them.
Team 10 includes one agent for up to 10 users, a dedicated server with a local model and two-factor login, updates, monitoring, backups, and 60 minutes of changes per month. It costs $1,990 setup plus $449 per month, net, with a 12-month minimum term.
On a dedicated server in your preferred region or on-premises. On-premises hardware and its requirements are discussed individually.
Your document storage, CRM or ERP, and internal APIs can be considered individually. We agree the scope and approve each connection; write actions need separate approval.
Yes. More users or agents, SSO, private networking, on-premises hardware, and custom integrations are discussed individually.
AI can make mistakes or miss context. Your team should review answers and sources before using them, especially for important decisions.
How we work
No surprises. Just clear conversations, honest scoping, and steady delivery.
We talk about your goals, constraints, and what success looks like. No pitch deck - just clarity.
I propose solutions that balance cost, complexity, and user experience. You decide what to build.
Clean, well-tested code shipped in tight increments. You see progress weekly, not at the end.
Confident deploys, monitoring, and handover.
Updates, maintenance and adjustments as your needs change. You keep one direct contact for support and advice.
$1,990 setup + $449/month
USD · net · 12-month minimum term
More users, agents, SSO, private network, on-prem hardware or custom integrations — let's talk.
The approach
Each agent runs in an isolated container on a dedicated server, in your preferred region or on-premises. The AI model and knowledge storage run locally in that environment. Your documents provide the context; roles and permissions determine access.
We define the use case and access rules together, document the data flows, and connect only the sources you approve. Your team reviews answers: AI can help find and prepare information, but people remain responsible for decisions.
Start with Team 10: one agent for up to ten users, with a predictable monthly price and no per-request billing. Larger environments are planned individually.
10+ years of professional full-stack development experience
Expert in modern TypeScript stacks, AI integration, and edge infrastructure
Performance- and UX-driven from the first sketch to the final deploy
I've built real-time, collaboration-heavy platforms like LEG.TJ, Contentoren, and Gruppenplan entirely from the ground up - product design, frontend, backend, and infrastructure.
My approach is product-driven: clarity, reliability, and long-term sustainability. Systems that scale gracefully and stay enjoyable to maintain. Strong in TypeScript, React, Solid, Node/Bun, Convex, and event-driven architectures.
For AI work, I focus on what actually delivers value - not demos.
Let's talk
Reach out with a sketch, a problem, or even just a question. I respond within a day, usually faster.
One direct point of contact for your questions and priorities.
Get advice before decisions and throughout operation.
Transparent documentation and a structured handover help you stay informed.
We maintain your system and apply updates regularly.
Keep exploring
Fast, accessible, on-brand
Marketing sites and landing pages built for performance and conversion. Static-first, SEO-ready, effortless to maintain.
Learn more
Complex products, built right
Thoughtful UX in collaboration tools - simple to use, seamless to navigate, and reliable as your team grows.
Learn more
Every conversation, one place
Bring your customer channels together, automate routine replies, and keep your team in control on your own server.
Learn more
Your company IT, on your own terms
Bring your essential company tools together on your own servers, independent of big-tech office suites. Choose only the apps your team needs.
Learn more