Application & AI control centre

Think It. Build It. Protect It.

Loglevel Sentinel gives you the tools to build, run and secure custom software.

It turns ideas into working business systems in minutes, built by your own people. It brings in the systems you already have — and keeps every one of them, and every AI request they make, under your control. On your own hardware, or in the cloud of your choice.

See how it works Log in to Sentinel

Minutes, not monthsBuild it yourselfImport your legacy systemsBake AI into old systemsEvery AI request on recordAI costs in real timeCost limits you setSecurity scored 0–100Risky prompts stopped before sendingSystems in Docker containersYour source codeNo licensing fees for systems you buildLocal and cloud AI modelsAccess to the latest frontier modelsSingle sign-onLive firewallLockdown in one move

What is Sentinel?

Every team needs its own customized system. Sentinel builds them — and keeps them all in one place.

Sales wants deal approvals. People wants onboarding. Operations wants shift planning. Hundreds of useful applications can quickly become a burden for IT. Sentinel is the control platform that makes them easy: it builds new systems, imports the ones you already have, runs each in its own container, and watches over all of them — including every question they send to AI.

Your systems

Knowledge libraryDocuments, answers, search

ComplianceRegisters, audits, reviews

OperationsPlanning, tasks, reports

Any new systemDescribed, built, deployed

Your existing systemsImported, converted, monitored

SentinelBuilds · monitors · protectsThink It. Build It. Protect It.

AI models

Local modelsOn your own hardware — private, no usage charge

Cloud modelsOnly when needed — keys never leave Sentinel

Build

Describe the system you need in plain words. Sentinel plans it, writes it, checks it and deploys it.

Monitor

See who is using what right now, how every system is doing, and what AI is being asked and what it costs.

Protect

A firewall that knows your systems, an AI gateway that scores every question, and lockdown when it matters.

01 Think It

Say what you need, in plain words.

A team needs a training dashboard, a compliance register or a smarter search. Write it the way you would explain it to a colleague. Sentinel turns the idea into a change package — a short, focused plan you can read before a single line is written.

Requests from users and administrators flow straight into the Builder
Big ideas are split into small, reviewable steps
Nothing changes until an administrator says go

Sentinel's Start a Change Package dialog with a new training dashboard task

02 Build It

Sentinel builds it, checks it and ships it.

In minutes, the Builder writes the code, runs the checks and deploys the result into the system's own container. Every change is saved as a snapshot you can roll back. Every file is readable in the codebase browser, and every database can be opened, explored and queried from one screen.

New systems start from templates and run in standard Docker containers
Git history, snapshots and a full audit trail for every change
Built-in database explorer with a SQL console — read-only by default

Sentinel's container overview with four running systems

Sentinel's database SQL console with a query and a bar chart of the results

03 Protect It

Then it watches over everything.

Who is signed in, which systems are busy, how fast they answer, what AI was asked and what it cost — live. The firewall shows where traffic comes from and keeps out what shouldn't reach you. And if something ever goes wrong, one deliberate move locks every system down.

Live requests, response times and people online, per system
Access per system: everyone, approved networks only, or closed
Lockdown with database dumps and a full incident record

Sentinel's firewall with a 3D globe of visitors and a live list of connections

Build and change it yourself

Minutes, not months.

A new system, a new field, a new report or a whole new module. The people who need it describe it in plain words, and Sentinel builds, checks and ships it while they wait. No development project, no waiting list, no consultants: your own team is in charge of its own systems.

The usual wayWeeks to months

Specify
Quote
Develop
Test
Go live

With SentinelMinutes

Describe it
Sentinel builds it
Live

Your own people build it

Anyone who knows the work can ask for a change in plain words. No programming, no developers, no tickets.

Change it when the business changes

A new field today, a new report tomorrow. Systems keep up with you instead of waiting for a supplier's next release.

Safe to try

An administrator approves with one click, and every change keeps a snapshot you can roll back in seconds.

No licensing fees

The systems you build are yours, source code included. No per-seat or per-system licence fees for anything you create.

Your existing systems

Hundreds of forgotten systems. Bring them all in.

A mid-sized company runs hundreds of smaller systems: a phone list, a booking tool, a supplier register that someone built years ago. Many are no longer maintained, and nobody watches them. Sentinel imports them, converts them to run in its own containers, bakes in AI where it helps, and from then on keeps them updated, monitored and secure, just like the systems it builds.

What you need: the original source code. Sentinel imports and converts each system from its source code, so you must have access to the original code for every system you bring in.

Before · unmaintained

Staff phone listLast changed in 2016 · no owner
Holiday requestsBuilt by a former employee
Equipment bookingOld framework · no updates
Supplier registerUnknown server · no backups
Quality reportsNo monitoring · errors unseen

After · in Sentinel

Staff phone listMonitored · up to dateAI searchHealthy
Holiday requestsMonitored · up to dateAI draftingHealthy
Equipment bookingMonitored · up to dateHealthy
Supplier registerMonitored · up to dateAI risk flagsHealthy
Quality reportsMonitored · up to dateAI summariesHealthy

New life, new abilities

Bake AI into systems that never had it.

Once a system is in Sentinel, you can add AI exactly where it helps the people who use it: search that understands questions, drafts, summaries or checks, without rebuilding the system. Every AI feature runs through Sentinel's AI gateway, so it is logged, scored for risk and kept within your cost limits.

Smart searchStaff phone list
Drafted repliesHoliday requests
Risk flagsSupplier register
SummariesQuality reports

Import

Bring in an existing system with its original source code and its data. Nothing has to be rebuilt from scratch.

Convert

Sentinel converts it to run in a standard Docker container, with backups and a snapshot of every version.

Monitor

Live traffic, errors, users, AI use and firewall rules, on the same screens as every other system.

Maintain

Updates, new features and AutoFix from the Builder. The original developer is no longer needed.

AI oversight

Know exactly how AI is used. Nothing reaches a model unseen.

Every AI request from every attached system passes through Sentinel's gateway. It is logged in full, scored for risk before it is sent, and linked to the person, the page and the model — with its cost counted in real time.

0%of AI requests logged — prompt, answer, tokens and cost

0–0security score on every question before it is sent

0+scores are stopped before a model ever sees them

0 clickto an AI Use Statement for auditors, as a PDF

Sentinel's AI Requests page with totals and a chart of requests by security category

Cost control

See the cost as it happens. Set the limits yourself.

Sentinel prices every AI request the moment it is made, so you always know what AI costs you: today, per system, per person, per page and per model. Not as a surprise at the end of the month.

Live spend per system, per model, per page and per person
A maximum cost per use on any AI page, set in dollars
Answers are shortened to stay under the limit, or the request is stopped before it is sent
Local models run at no usage charge and are never limited

Live AI spend · today$149.61

Maximum cost per use$1.00

Illustration · names and values are examples

⌁

Every request, on record

The prompt, the complete answer, tokens, cost and who asked. Search it, filter it, export it to Excel.

◎

Scored before it is sent

An automatic reviewer rates each question from safe to dangerous — prompt injection, secrets, personal data, off-topic use.

⛨

Stopped when it matters

With the pre-send check on, requests that score 60 or more never reach a model. Your own review always has the last word.

▦

The right AI for the right people

See every page that uses AI, which user groups can open it and which models answer there.

Costs, live and limited

Spend per system, per model, per page and per person as it happens, with a maximum cost per use wherever you want one.

⌂

Local first

Open-weight models run on your own hardware at no usage charge. Cloud models are used only when needed.

The AI Use matrix showing which user groups can use which AI pages and models

Models in use with their cost, and the security review and controls

Bring your own AI

Your own AI, plugged into Sentinel. Still under your rules.

Developers keep the AI they already work with — Claude Code or Codex on their own computer — and connect it to Sentinel with one pasted text. AI agents that run inside your network can be locked into an NVIDIA OpenShell sandbox. Either way, every AI call passes through Sentinel, and Sentinel itself stays out of reach.

People's own Claude Code and Codex connected to Sentinel over MCP next to the attached systems; Sentinel's AI gateway sends every request to hosted AI (Claude, GPT), on-prem AI (Gemini, free) and an agent inside an NVIDIA OpenShell sandbox

Illustration · names and values are examples

Enterprise

Your own AI, connected in a minute

Claude Code or Codex talks to Sentinel over MCP, with a personal token. It sees what you see — and nothing more.

Click the plug icon in the PlannerSentinel makes a personal, revocable token
Paste one text into Claude Code or CodexIt sets up the connection on its own
Ask away“Why does the import in OpgaveKlar fail?”

Reads the code of every system you have access to
Asks Sentinel and gets its answers — brief, vulnerabilities, live status
Container logs with secrets masked, files as one-time downloads
Read-only, every call logged and rate-limited — Sentinel's own code is off-limits
On the office network it uses the local address automatically

Enterprise · option

NVIDIA OpenShell, when an agent needs a cage

OpenShell is NVIDIA's open-source sandbox runtime for AI agents. Sentinel integrates it as an option: each sandbox runs in its own micro-VM, and a policy decides exactly which networks and files the agent can reach.

Network: allow-listFiles: own workspaceNon-rootEvery action audited

Every policy is proven against your Sentinel boundary before it is saved
A built-in isolation test: non-root, system folders blocked, cloud metadata refused
Sandboxes, policies and versions in one view under Security
Runs on your own hardware, with telemetry switched off

Live control

See everything as it happens.

Sentinel's Live page follows every request through the gateway in real time: where it came from, which system it went to, how long it took, and how risky the AI part was. Dot speed follows response time — you can literally watch your systems breathe.

Sentinel's Live page: requests flowing from sources through the Sentinel gateway to systems and AI

Last requests per system, people online and the live request stream

Sentinel's dashboard with service health and activity

Development and error correction

Systems that keep getting better — and fix themselves when they break.

Round after roundAlways
improving

One-click development

From idea to update, again and again.

1A user requests a featureFrom inside the system they use
2An administrator starts AutoBuildOne click, after a quick read
3Sentinel builds the featureCode, checks and deployment
4The application is updatedWith a snapshot to roll back to

Round after roundSelf-
healing

Real-time error correction Enterprise

From bug to fix, automatically.

1A bug is detectedIn any attached system
2Sentinel detects and diagnoses itLogs, code and recent changes
3Sentinel runs AutoFixA focused, reviewable repair
4The application works againRecorded in the change history

Why Sentinel

Your systems. Your source code. Your hardware, or your cloud.

Runs wherever you choose

Your own servers
AWS
Microsoft Azure
Any other hosted platform

⬡

Docker deployment

Every system runs in standard Docker containers, on your own servers or in AWS, Azure or any other hosted platform. Backed up, restorable and recoverable on its own.

✦

Your plugins

You own every plugin you create, including its source code, and can reuse it across systems.

❖

Your systems

You own every system you develop. Nothing is locked inside someone else's platform.

No licensing fees

No per-seat or per-system licences for the plugins and systems you create.

AI security by design

Every AI request from every system passes through Sentinel, is logged in full, risk-scored and reaches a model only when it is safe.

User-aware, context-aware AI

Sentinel knows who is asking, from which system and module — and with what data — so answers and access follow the person.

One-click development

Describe a change. Sentinel's AI builds it in the right module, with cost, history and rollback on every step.

One-click deployment

Accepted changes go live in the system's container automatically, with health checks and instant recovery.

AI value alignment

Your company's values, context and rules shape how every request is judged — from safe to dangerous.

Editions

Two editions. Start with Essentials, grow into Enterprise.

Both editions build, monitor and protect your systems, and in both every AI request goes through the same gateway: logged, scored for risk and kept within your cost limits. Enterprise adds scale, self-healing, compliance tools and high availability.

Sentinel Essentials

For a team or a department getting started with its own systems.

Up to 3 systems, plus 1 imported existing system
The Builder and one-click development
AI gateway, Live monitoring and firewall
30 days of history · standard support

For the whole organisation

Sentinel Enterprise

For companies with many systems, strict compliance needs and no room for downtime.

Unlimited systems, imports and administrators
AutoFix self-healing and parallel builds
Compliance tools, single sign-on and AI value alignment
Several hosts or clouds with high availability · SLA

Essentials and Enterprise compared
Feature	Essentials	Enterprise
Systems
Systems (Docker containers)	Up to 3	Unlimited
Existing systems imported and converted	1	Unlimited
Administrators	2	Unlimited
Building
Builder and one-click development	✓	✓
Builds	One at a time, with a monthly quota of change packages	Several at once, no quota
AutoFix — self-healing error correction	—	✓
Bring your own AI — your local Claude Code or Codex, plugged into Sentinel	—	✓
AI and security
AI gateway: every request logged, risk-scored and cost-limited	✓	✓
Live monitoring and firewall	✓	✓
AI value alignment with your own company policies	—	✓
Single sign-on	—	✓
OpenShell integrated — NVIDIA's sandbox runtime for AI agents	—	✓
Daily brief — your code checked against vulnerability databases, plus the day’s news, as a PDF newsletter	—	✓
History and compliance
AI request logs, Live history, snapshots and backups	30 days	Unlimited
One-click AI Use Statement PDF for auditors	—	✓
Lockdown with database dumps and a full incident record	—	✓
Hosting and support
Hosting	A single host	Several hosts or clouds, with high availability
Support	Standard support	SLA
Licensing fees for the systems and plugins you build	None	None

Not sure which edition fits? Tell us about your organisation and we will help you plan how to bring Sentinel in.

Live requests per system — Live requests

Think It. Build It. Protect It.

One control centre for your business applications and the AI inside them — built with you, running on your hardware or in your cloud, watched over every second.

For Sentinel administrators. Screens on this page come from the running system; names and addresses have been replaced.