LoopBack
Your company’s own AI.
Secure, compliant and EU-hosted: the complete AI platform, behind one gateway. Smart routing, the AI Firewall, your own Private AI and agents, all on your own server. Your data stays with you, and you pay for the server, not per token.
Built for management and IT: more AI per euro, security and compliance from day one, and no lock-in.
Your employees already use AI. Just not yours.
Your employees use AI whether it is allowed or not. The result: sensitive data outside your control, token costs IT cannot predict, staff who run out of tokens mid-task, and Claude or OpenAI plans too restrictive to roll out. And you are not ready for the EU AI Act. LoopBack brings it all onto one platform you own, so AI becomes a managed tool instead of a shadow risk.
Use cases by team
One platform. Every department. Faster from day one.
Sales
Offers and customer insight grounded in your CRM, ready in minutes instead of hours.
Support
Faster answers built on your own docs and past cases, not the open web.
Marketing
Content written in your own tone, drawing on your material.
HR
Policy and onboarding answers straight from your knowledge base.
Finance
Reports and status pulled from your ERP in plain language.
Ops & IT
Code and automation without source code leaving the building.
What we offer
LoopBack isn’t one tool but a platform of five services that share the same gateway, the same login and the same controls. Smart routing and cost optimization, security and compliance, Private AI, agents, and bespoke AI built for your business.
How many paid parental-leave days do we offer in Denmark?
Your handbook grants 24 weeks of paid parental leave in Denmark, shared between parents.
Employee-Handbook.pdf · p.14EU-hosted · data never leaves
Your own AI on your own server
Private AI
Your own Private AI, on your own server, for the whole company. You pay for the server, not per token, so there are no token quotas and no rate limits, and staff never run out mid-task. The everyday chat and the code assistant are both capabilities under Private AI, with your own rules for what can be shared.
- ▸Draft emails, notes and documents
- ▸Summarise long texts and meetings
- ▸Research and quick answers
- ▸A code assistant in the tools your developers already use
Question
What did we agree on pricing with Nordwind in Q2?
Smart Routing
You agreed a 12% volume discount above 500 seats, valid through 2026, with a quarterly review.
Sources
- Nordwind-MSA.pdf · p.7
- pricing-2026.xlsx · Sheet 2
- CRM · Nordwind A/S
More AI per euro
Smart Routing & Cost Optimization
Each request is routed to the model that solves it best and cheapest: simple questions to cheap models, only the complex ones to the frontier, so you get more AI per euro. A memory layer stores your code, documents and decisions locally and reuses the context instead of resending it, grounded in your CRM, ERP and documents.
- ▸Route each request to the best, cheapest model
- ▸Reuse context from your own memory layer
- ▸Answers grounded in your own data and documents
- ▸Sales support: customer insight and next steps
| Agent | Trigger | Last run | Status |
|---|---|---|---|
| Invoice intake | New email | 2 min ago | Done |
| Lead routing | Form submit | 14 min ago | Running |
| Weekly report | Mon 08:00 | Scheduled | Queued |
Invoice intake · run
- ✓ Read
- ✓ Extract
- ✓ Post to ERP
- 4 Notify
Agents that handle the routine work
AI Agents
Agents that work across your systems and handle the routine work around the clock, 24/7. An invoice lands in the inbox, the agent reads it, pulls out the fields and forwards them to your ERP without anyone touching it.
- ▸Automate invoice and document flows
- ▸Lead routing and follow-up
- ▸Generate reports automatically
- ▸Email triage and categorisation
AI suggestion
Tab to accept
Solutions built for your business
Bespoke AI
Solutions built for your business, on your own platform. Custom AI apps, RAG over your docs, fine-tuned and purpose-built models, an AI Employee, and customer-facing bots, all shipped to production by the team that runs the platform and all behind your own gateway. The code assistant ships as a capability under Private AI; here we build the AI that is unique to you.
- ▸Custom AI apps built for a specific job
- ▸RAG over your documents, contracts and handbooks
- ▸Fine-tuning on your data and purpose-built models
- ▸An AI Employee in Slack or Teams
Requests today
12,480
PII redacted
318
Budget used
62%
| Team | Model | Data | Status |
|---|---|---|---|
| Sales | Claude | CRM (grounded) | Allowed |
| Support | EU-hosted | Helpdesk | Allowed |
| Finance | Self-hosted | On-prem only | Restricted |
The AI Firewall
AI Security & Compliance
All AI traffic passes through one control point on your own server. Guardrails check every call before, during and after, PII like CPR numbers, emails and card numbers is masked or blocked, prompt injection and moderation are caught, and every request is logged. GDPR and the EU AI Act, from day one.
- ▸One control point for all AI traffic and keys
- ▸PII filtering: CPR, email and card numbers masked or blocked
- ▸Prompt-injection and moderation checks
- ▸Per-team and per-key access rules
Integrations
OpenAI-compatible: works with the tools you already use
LoopBack speaks an OpenAI-compatible API, so it drops into your dev tools (Cursor, Cline, Continue, Aider, Open WebUI, n8n), SaaS with bring-your-own-LLM (Salesforce, HubSpot, Notion, Slack), and the included clients (Desktop, Web Chat, Code CLI, Agents), grounded in your own data rather than the open web.
Missing your system? We connect to most via API.
How it works
01 / 03
Everything goes through the AI Firewall
Whether it’s chat, code, agents or internal apps, every request passes through one control point on your own server. Guardrails check each call before, during and after, PII is filtered, prompt injection and moderation are caught, and everything is logged for the EU AI Act. One point of control over all your AI.
02 / 03
A memory layer that stays with you
LoopBack stores your code, documents and decisions locally and reuses that context instead of resending it, which saves tokens, time and frustration. Over time it learns your customers, processes and deals, and answers get sharper without knowledge ever leaving your servers.
03 / 03
The right model for the job
Why not just ChatGPT or Copilot?
Public AI tools are built for individuals, not for running a company.
Why your own AI, and why now
Your team already uses AI. The question is whether you’ll govern it.
You pay for the server, not per token
Your own Private AI runs on a fixed price: a monthly subscription when hosted, or upfront on-premise. No token quotas, no rate limits, so staff never run out mid-task.
Audit-ready from day one
EU-hosted or on-premise, GDPR-aligned and with full logging for the EU AI Act. Not something you bolt on later.
Never locked to one vendor
Switch or mix models freely as the field moves, without changing your apps. Export data and configuration in open formats.
One bill, no ops
We host, update and monitor the platform for you. You use it, we keep it running.
The EU AI Act is phasing in now, and your employees are already using AI outside your control. The longer you wait, the more data is already out of the building.
Built for IT to say yes
Every employee gets AI. IT keeps control.
One gateway, one view
All AI traffic passes one control point with SSO, role-based access and per-team keys.
Hosted on your terms
On your own servers, or in an EU data center in Denmark or Germany. Your data never trains anyone else’s models.
Logging and guardrails
PII is filtered automatically. Every request is logged for the EU AI Act. Budgets and rate limits are set per team.
No lock-in
Switch models without app changes. Export data and configuration in open formats.
The right model for the job
Frontier
Claude, GPT and the like. Maximum quality for complex tasks.
Standard
Fast, cheap models for high-volume tasks.
EU-hosted open
GLM, MiniMax, DeepSeek and Kimi. Claude-class quality with EU data residency.
Self-hosted
Open-source models on your own GPUs. Data never leaves your servers.
Your own Private AI runs EU-hosted or on-premise: you pay for the server, not per token. Managed frontier models (Claude, OpenAI) stay optional and external, for the few tasks that need them. You move freely between tiers with no app changes.
What we promise
Time savings
Routine work that takes hours today is handled in minutes.
Security and compliance
GDPR and the EU AI Act are built in, not bolted on afterward.
Better decisions
AI grounded in your own data, not generic guesses.
Cost control
One invoice and predictable usage. You pay for the server, not per token.
Flexibility
Switch models freely, with no lock-in and no changes to your apps.
Scalability
Add employees and services without re-architecting.
Choose the level that fits
Three packages for companies with 10–250 employees. Start small and expand as the need grows.
AI for everyone
Starter
from €690/mo
AI Security & Compliance + Private AI
- ✓Private AI for the whole team
- ✓GDPR and EU AI Act ready
- ✓Up to ~50 employees
AI that knows your business
Growth
from €3,790/mo
Security + Private AI + Smart Routing & Cost Optimization
- ✓AI connected to your CRM, ERP and documents
- ✓Coding help in your IDE
- ✓51–150 employees
AI that runs your busywork
Enterprise
from €9,490/mo
All five services
- ✓Full platform incl. AI Agents and Bespoke AI
- ✓On-prem or EU data residency
- ✓151–250 employees
Starting prices per month, ex. VAT. Platform fee and usage billed on top. One-time setup from €1,500.
See your price in under a minute
Pick services and number of employees, and get a starting estimate right away.
What do you need?
AI Security & Compliance is the foundation and always included.
How many employees?
Which systems should LoopBack connect to?
Pick the systems your AI should have access to (optional).
Anything else?
Your estimate
Recommended plan
Starter
+ usage
One-time setup, from €1,500
Includes
More than 250 employees?
We’ll put together a tailored plan. Tell us a bit about your needs and we’ll get back to you.
We’ll send a tailored quote within one business day, no obligation.
Starting estimate based on our public starting prices, ex. VAT. Your final quote depends on number of employees, usage and integrations.
Frequently asked questions
How is LoopBack different from ChatGPT or Copilot? +
ChatGPT, Copilot and the like send your data out of the building, don’t know your business, and lock you to one vendor. LoopBack runs on your own server or in EU hosting behind your VPN, grounds answers in your own systems, and lets you switch models freely.
What happens to our data? +
It stays with you. LoopBack runs either on-premise or in an EU data center in Denmark or Germany, and your data is never used to train anyone else’s models. Every request is logged for your own audit and EU AI Act compliance.
Do we have to run anything ourselves? +
No. We host, update and monitor the platform for you. If you’d rather run it on-premise, we handle that setup too.
What if we want to switch away? +
You export your data and configuration in open formats and take it with you. Models are swappable, so there’s no re-platforming.
How long does it take to get started? +
We start with a workshop, run a pilot with one team, then roll out to the whole company. Typically live within a few weeks.
Up and running in a few weeks
Workshop
We map your needs, systems and compliance requirements together with you.
Pilot
LoopBack is set up with your integrations, and one team starts using it.
Rollout
The rest of the company comes on, with training and ongoing support.
Onboarding and setup is €1,500 in 9 out of 10 cases, incl. integrations, loading your documents and training. Excl. hardware and server/hosting costs. For special needs you get the price before we start.
Let us give your company its own AI
See it live in a 30-minute demo, get a quote tailored to your size, or start a pilot that is running within weeks.