Deployment

Runs where your data lives

Use Aggrelis as a hosted service, on an instance only for you, or in your own cloud or data center. The same engine and the same APIs in all three.

Talk to us Request access

Illustrative

Three ways to run it

Pick the model that fits

Start on the hosted service, get an instance of your own, or run Aggrelis in your own infrastructure. In all three, every workspace keeps its own database.

The EU at launchMore regions to come

Hosted, shared

Subscribe and start

Your workspace on our shared instance in the EU, kept apart from every other customer in a database of its own.

One workspace per subscription
We run, update, scale and back it up
Limits come from your plan

Request access

Dedicated

An instance only for you

An instance that takes only your workspaces, run by us, with as many workspaces as you need: per team, brand, country or customer.

Nothing shared with other companies
A location chosen with you
Limits and retention set for you

Talk to us

Self-hosted

Run it yourself

The whole engine in your own cloud or data center, right where your data already lives. Deploy it with Helm or Docker Compose.

Both consoles and every API
Monitoring, alerts and runbooks included
A license that never calls home

Talk to us

Side by side

What changes between them

The engine is the same. What differs is who runs it, where your data lives and where your limits come from.

Hosted, dedicated and self-hosted compared
	Hosted, shared	Dedicated	Self-hosted
Who runs it	We do: upgrades, scaling, monitoring and backups	We do, on an instance that takes only your workspaces	You do, with our Helm chart, compose files, alerts and runbooks
Where your data lives	The EU at launch, more regions to come	A location we run, chosen with you	Wherever you run it. It leaves only through what you set up, such as webhooks
Limits come from	Your plan	Your contract	Your license and your installation's settings
AI modelsFor decisions and AI help	Our model catalog, run in the EU	Our catalog, in the instance's region	Your own connections: Azure OpenAI, Vertex, Bedrock, any OpenAI-compatible or self-hosted model
Signing inTo the console	Passwords, passkeys, two factors, and your own single sign-on	The same	The same, with an identity service shipped beside the engine
System consoleWorkspaces, servers, AI connections	Ours	Ours	Yours
How you get it	Sign up and check out	Talk to us	Talk to us; we issue a signed license file

Self-hosted

One image, in your infrastructure

The whole engine, both consoles and every API ship as one container image. Run it on Kubernetes with the Helm chart, or on Docker hosts and Docker Swarm with the production compose files, on your own PostgreSQL and Redis.

Helm chart with autoscaling, network policies and a Grafana dashboard
Production compose files: secrets as files, read-only containers
Scale ingest, queries and admin on their own
Try the full stack on a laptop with one command

Read the self-hosting docs

Kubernetes 1.25 or later

# API, worker, provisioner, migrations,
# network policies, alerts and a dashboard
helm install aggengine oci://ghcr.io/bjeber/charts/aggregation-engine \
  --version <version> \
  --namespace aggengine --create-namespace \
  --set secrets.existingSecret=aggengine \
  --set config.AE_REDIS_ADDR=redis-master.redis:6379

Docker hosts and Docker Swarm

# deploy/compose in the release
cd deploy/compose
cp .env.example .env    # image, Redis address, authentication
docker compose --env-file .env up -d
docker compose ps

values.yaml

metrics:
  serviceMonitor:
    enabled: true
  prometheusRule:
    enabled: true    # alert rules, each with a runbook
  grafanaDashboard:
    enabled: true

Shortened from the self-hosting docs.

Ready to operate

Ships with what your operations team needs

Dashboards, alerts and runbooks come with it, so you run Aggrelis like the rest of your stack.

Alerts with runbooks
More than 20 alert rules, each with a runbook: backlogs, webhooks, memory, sign-in failures, the license and more.
Metrics, traces and logs
Prometheus metrics, OpenTelemetry traces, JSON logs and a ready-made Grafana dashboard.
Redis rebuilds from PostgreSQL
PostgreSQL is the source of truth, with point-in-time recovery. Redis needs no backup of its own.
Upgrades that migrate first
Migrations run before every upgrade, and each workspace's schema is migrated on its own.
Sign-in included
An identity service ships beside the engine, with a break-glass operator for when it is down.
Signed images
Signed container images with a software bill of materials and provenance.

Licenses

A license that never stops production

A self-hosted installation runs under a signed license file. Every instance checks it offline, at startup and every hour. It never calls home, and a renewed file applies without a restart.

Sets workspaces, events per month and features
Can be bound to one installation
If it lapses, ingest, queries, reads and sign-in keep working
Alerts for 30 days before and after expiry; only then are admin changes refused

Talk to us about a license

Illustrative

Dedicated or self-hosted

Tell us where it needs to run

We go through it with you: the location, the volume and who operates it. There's no public offer for these yet, so we agree the details together.

Talk to us Request access