Self-hosted · Apache-2.0
A self-hosted recommendation engine you run on your own infrastructure
KaireonAI is an open-source, Apache-2.0 recommendation and decisioning engine. Deploy it as a single application on your own servers, score with nine ML engines, and return the best action in a real-time API call — with full control of your data.
Why self-host your recommendation engine
A hosted recommendation service means your customer data leaves your environment and your decisioning logic lives in a black box you cannot inspect. Self-hosting removes both problems.
Your data stays yours
Run the engine on your own Linux, Docker, or Kubernetes environment and keep customer data inside your own environment. No third party sits between you and your customers, and nothing is metered per decision.
One thing to deploy
KaireonAI ships as a single application that serves both the UI and the decision APIs, backed by PostgreSQL. Self-hosting is one deploy, not a fleet of services to wire together.
Open and inspectable
Apache 2.0 licensed and self-hostable, so the pipeline stages, the models, and the decision traces are visible rather than hidden behind a service. You can understand and control every recommendation.
More than a recommender — a decisioning engine
The item a person is most likely to click is not always the one you are permitted to show them, the one that serves the business, or the one that respects fatigue rules. A self-hosted recommendation system that only ranks by predicted engagement misses all of that.
KaireonAI runs a full decision pipeline on every request: inventory, eligibility and fit filters, contact policy, match scoring, and multi-objective ranking. Eligibility, policy, scoring, and ranking all execute per call, and the winning action comes back through a real-time Recommend API. The customer response flows back through a Respond API, closing the loop so the next decision is a little smarter.
Nine scoring engines, built in
Scoring is a choice, not a single algorithm. Assign any of nine engines to a decision — each scores inside the pipeline, with no external inference hop. The bandit and online-learning options matter for cold starts, where a model that explores deliberately learns which actions work far faster than one that waits for a scheduled retrain.
Secure by default
Self-hosting only helps if the engine is safe to run. KaireonAI enables per-tenant Row-Level Security automatically, encrypts data at rest with AES-256-GCM, and enforces RBAC with a hash-chain audit trail. Deploy it on your infrastructure with full control over who can see and change what.
Frequently asked questions
What is a self-hosted recommendation engine?
A self-hosted recommendation engine runs on infrastructure you control rather than a vendor's cloud. With KaireonAI you deploy a single Apache-2.0 application on your own Linux, Docker, or Kubernetes environment, so customer data and the recommendation logic stay inside your own environment.
Is KaireonAI a recommender or a decisioning engine?
Both. A recommender answers what a person is likely to engage with. KaireonAI goes further and answers what single action you are allowed to take and should take given eligibility, contact policy, and business value — so the winner is genuinely best, not just the most clickable.
What models can the engine use?
Nine scoring engines are built in: Scorecard, Bayesian / Naive Bayes, Logistic Regression, Gradient Boosted Trees, Thompson Bandit, Epsilon-Greedy, Neural Collaborative Filtering, Online Learner, and an External Endpoint to call your own model. Each scores inside the pipeline, per request.
How does it get my data?
50+ data connectors bring data from object stores, streaming platforms, warehouses, lakehouses, CRMs, relational databases, and REST APIs. Schema-driven ingestion creates real PostgreSQL tables, and customer profiles can be enriched at decision time.
Keep reading: What is Next-Best-Action? · Personalization without PII · Platform features
Deploy it where your data lives
Explore the live playground, then self-host the engine on your own infrastructure.