Features - Everything you need to run copy experiments

From idea to live experiment in minutes. Xaiku handles generation, analysis, prediction, and measurement in one tightly integrated workflow.

AI Variant Generation

Describe your goal in plain language and Xaiku builds a fully structured template — fields, names, and character limits — extracted directly from the conversation. No forms to fill out, no configuration to wrestle with.

Each variant produced is structurally distinct: different angles, different emotional hooks, different calls to action. You get genuine creative diversity instead of the same sentence rephrased five ways.

Configure exactly how many variants you need, from a quick pair to a full batch of ten. Every run respects your voice profile, so the output sounds like you from the first draft.

Phrase-Level Analysis

Xaiku breaks every variant into semantic phrases and tags each one with one to three emotions — curiosity, urgency, trust, and more. Color-coded visualisation makes the emotional arc of your copy immediately visible.

Each phrase also carries a gene type classification drawn from five categories that describe the rhetorical function it serves. Understanding why a phrase works is as important as knowing that it does.

The three-tab phrase editor lets you inspect raw text, emotion tags, and gene types side by side. Edit any phrase in place and watch the analysis update in real time.

Voice Profiles

A voice profile captures the personality behind your brand. Toggle eight text style dimensions — headline, storytelling, active voice, first-person, factual, simple, concise, and personalised — to define exactly how your copy should sound.

Paste in accepted and avoided excerpts to give the AI concrete examples of on-brand and off-brand language. The toolkit layer turns those examples into enforceable rules applied during every generation run.

Voice profiles persist across experiments and can be shared across your organisation. The settings hierarchy means experiment-level overrides are always respected without losing the baseline brand voice.

AI Forecast

Before a single impression is served, Xaiku scores every variant across five dimensions: readability, relevance, urgency, specificity, and trust. The composite score gives you a performance prediction grounded in linguistic signals rather than gut feel.

The leaderboard view ranks all variants in an experiment so you can prioritise which ones to test first or skip entirely. Radar charts show the dimensional breakdown for each variant at a glance.

Forecast scores update automatically whenever you edit a variant or add new ones. The prediction model is version-controlled alongside the rest of the generation system, so improvements roll out transparently.

SDK Integration

One snippet connects your front end to the experiment engine. The SDK supports four frameworks — vanilla JS, React, Next.js, and Vue — with identical APIs across all of them. Switch environments without rewriting integration code.

Test mode mirrors live behaviour without recording real events, so you can validate the integration before launch. Automatic distribution handles traffic splitting based on the weights you set in the dashboard.

Event tracking is built in. Every impression and conversion is captured and forwarded to the analytics pipeline the moment it happens. Full documentation is available in the docs.

Analytics

Results are validated with statistical significance testing, not just raw conversion counts. The dashboard surfaces confidence levels alongside lift percentages so you know when a winner is real and when you need more data.

Variant performance charts, time-series views, and interaction heatmaps give you multiple lenses on the same experiment. Sankey diagrams trace the full user journey from first impression to final conversion.

Radar charts overlay the forecast scores against actual performance, closing the loop between prediction and reality. The gap between the two is a learning signal for every future experiment.

Ready to see it in action?

Start your first experiment in minutes. Plans start at €8/mo.

Start experimenting