The decisioning system
for teams shipping at scale.

Production-grade bandit infrastructure. Built for teams that learn from every request.

01 / 04Homeevery experiment, one screen — live throughput, allocation, top-arm reward.
BUILT ON THE STACK YOU ALREADY RUN
PythonRedisgRPCPostgreSQLKubernetesFastAPIClickHouse
ADAPTIVE

Watch how qbrix optimizes your site in real-time.

Every request shifts traffic toward what's working. When the world changes, qbrix adapts — no dashboards to babysit, no tests to declare.

Exploring

qbrix distributes traffic equally across all variants, collecting initial performance signals.

Exploiting

Shift detected

Re-exploring

New winner

PRODUCT

Built for production.
Designed to disappear.

Drop qbrix into your stack and forget it's there. Kubernetes-native, multi-tenant, and self-optimizing from request one.

REAL-TIME ADAPTIVE

Real-time adaptive

Traffic moves to what's working. No manual analysis or waiting for significance — self-adjusting in real minutes.

SELF-OPTIMIZING

Self-optimizing

qbrix picks which policy fits your data — and switches when conditions change.

BetaTSstochastic
selected
UCB1Tunedstochastic
LinUCBcontextual
MULTI-TENANT & RBAC

Multi-tenant & RBAC

Full isolation between tenants, three-role RBAC, and a bad audit trail — run it out-of-the-box.

Adminscoped
Memberscoped
Viewerscoped
PRODUCTION SCALE

Production scale

3.2 msavg selection latency
50kreq/s
10 msp99
FEATURE GATES

Feature gates

Progressive rollouts, targeted segments, and accelerate feature — all in one SDK.

new-checkout-flow
ai-recommendations
dark-mode-v2
OBSERVABILITY
LIVE

Every decision, in the open.

Stream selections, rewards, and policy switches as they happen — no black-box ML, no waiting for nightly reports.

TODAY — MAY 12
12:42:18PM
selection.servedarm: pro_plan · exp: pricing_v2
12:42:14PM
feedback.receivedreward: 1.0 · arm: checkout_b
12:42:11PM
selection.servedarm: variant_a · exp: onboarding
12:42:07PM
experiment.updatedexperiment · exp_feed · by: admin
CONTEXTUAL

Personalized at request time

Per-user features choose the best arm. New visitors and power users see what works for them — not what works on average.

CONTEXT
device:mobileplan:proregion:EU
LINUCB
variant_b
87%
DEVELOPER EXPERIENCE

Integrate in minutes.

A handful of Python calls is all it takes to stand up a production-grade bandit system.

app.py
1from qbrix import Qbrix
2
3client = Qbrix(api_key="optiq_...")
4
5experiment = client.experiment.create(
6 name="checkout-banner",
7 pool_id=pool.id,
8 policy="auto", # meta-bandit portfolio
9 policy_params={"reward_type": "binary"},
10)
Drop-in Python SDK

pip install qbrix — one import, one API key, and you're in production.

Sensible defaults

The meta-bandit portfolio selects and tunes the policy for you. Opt-in to advanced control when ready.

REST or gRPC

Full HTTP REST API and gRPC stubs — pick what fits your stack.

View API reference
USE CASES

One platform.
Every decision.

Wherever you choose between variants — content, products, prices — qbrix learns what works and adapts in real time.

Content & UX

CONVERSION

Which headline converts best? qbrix continuously tests and shifts traffic to the winning variant — no manual analysis, no waiting for significance.

Headlines, landing pages, onboarding flows, CTAs

Recommendations

PERSONALIZATION

Show each user the item most likely to convert — based on real-time signals, not static rules. Contextual bandits factor in who the user is.

Product feeds, search ranking, featured items

Dynamic Pricing

REVENUE

Find the price point that maximizes revenue, not just clicks. qbrix handles the exploration-exploitation tradeoff automatically.

Price testing, discount tiers, subscription offers

Feature Rollouts

RELEASE

Go beyond random percentages. qbrix learns which segments respond best and progressively expands exposure — maximizing adoption.

Progressive rollouts, beta targeting, adoption lift

Campaign Optimization

MARKETING

Which creative drives the most signups? Let qbrix allocate budget toward what works — automatically, in real time.

Ad creatives, email campaigns, channel allocation

Search & Ranking

ENGAGEMENT

Optimize the order of results, products, or content. qbrix learns which ranking drives the most engagement per user context.

Search results, product listings, category ordering

PRICING

Simple. Transparent.
Free to start.

Get started for free. Scale as you grow. No hidden fees, no surprises.

Free
€0

For side projects and experimentation. Get started in minutes.

Up to 5 seats
100 requests / minute
2 API keys
3 active experiments
Full algorithm suite
Feature gates
Community support
Get started
Starter
€49/mo

For early-stage teams scaling beyond side projects.

Everything in Free, plus:
Up to 15 seats
1,000 requests / minute
5 API keys
15 active experiments
Scheduling rules
Workspace invites
Email support
Start with Starter
ProMOST POPULAR
€149/mo

For growing teams. Everything you need to run experiments in production.

Everything in Starter, plus:
Unlimited seats
10,000 requests / minute
25 API keys
Unlimited experiments
Insights dashboard
Priority support
Upgrade to Pro
Enterprise
Custom

For organizations that need full control, audit trails, and SLAs.

Everything in Pro, plus:
Unlimited everything
RBAC (roles & scopes)
Event log & audit trail
Selection & feedback tracing
Dedicated support + SLA
Contact sales
GET STARTED

Ship smarter decisions
in minutes.

Connect your app in minutes with our Python SDK or REST API. Sign up free and start running your first experiment today.

✓ No credit card required✓ Free forever tier✓ Self-host or cloud