Stop guessing what converts.
Bandit knows.
Drop in the SDK, wire your rewards, let it run. Real bandit algorithms find what works on your funnel — and AI generates new variants when results plateau.
Traffic automatically shifts to the best-performing variant. No manual intervention needed.
Built for teams that ship fast
Find what converts. Without rule trees.
Most teams are stuck guessing — or maintaining segments and rules that rotted six months ago. Bandit lets the math figure it out.
- Segments and rules someone hand-coded once
- Doesn't learn — keeps serving what was best a year ago
- You still don't know what actually converts
- Real bandit algorithms converge on what works
- AI generates new variants when results plateau
- Joint funnel optimization — the whole chain learns together
What Bandit does for your funnel
Real bandit algorithms, AI variant generation, and joint funnel optimization — wired together so the math converges on what converts.
Real bandit algorithms
Thompson Sampling, UCB1, Epsilon-Greedy, and Linear Contextual — running on your funnel, balancing exploration and exploitation on every request.
Sub-100ms assignments
Edge-deployed on Cloudflare Workers. Bandit picks the next variant fast enough to render inline — no flicker, no fallback.
Drop-in SDK
TypeScript SDK, REST API, edge-deployed. Install, call getAssignment, wire your reward. Honest about the work — not a magic wand.
AI variants when you plateau
When the algorithm runs out of meaningful lift on your existing variants, AI generates new ones from your performance data. The system never stops finding edge.
Joint funnel optimization
Funnels are optimized as one system, not four loose experiments. A copy change at step 1 informs the math at step 4 — Bandit converges on the funnel-optimal path.
How Bandit stacks up
Most teams trying to optimize a funnel pick one of these four paths. Bandit is the only one that actually learns.
Three steps. Then hands-off.
From integration to convergence in minutes. After that, the math runs itself.
Drop in the SDK
Install the package, initialize with your API key, and call getAssignment() wherever a decision lives. Minutes, not days.
Wire your rewards
Define what counts as a win — conversion, click, revenue. Pick an algorithm. Add variants yourself, or let AI generate them.
Let it run
Bandit converges on what works. New variants get generated when results plateau. You get a real-time dashboard and a system that keeps finding lift.
Five lines. The math takes it from there.
The SDK handles assignment caching, event batching, and automatic retries. You render the variant and fire the reward — Bandit converges on what wins.
import { BanditClient } from '@runbandit/sdk';
const bandit = new BanditClient({
apiUrl: 'https://runbandit.com',
apiKey: 'your-api-key'
});
// Get the best variant for this user
const { treatment } = await bandit.getAssignment('headline-test', 'user-123');
// Track when they convert
bandit.trackEvent({ eventType: 'CONVERSION', value: 29.99 });
Stop guessing. Start converging.
Drop in the SDK, wire your rewards, and let the math find what converts. Free tier available, no credit card required.