aiva - measures and steers your visibility in LLMs
aiva is not a self-service dashboard, but a consulting-backed visibility analysis for globally operating industrial companies. aiva measures where you appear in your buyers’ AI answers and against whom you lose per industry and market - across every relevant LLM, Western and Chinese, in all markets and languages, China included - and delivers the measures along with it.
Features
All models - Western and Chinese
All Western frontier models including Claude - plus the complete Chinese LLM ecosystem: DeepSeek, Doubao, Ernie, Qwen, Hunyuan and Kimi, covered via our Shanghai infrastructure. No other tool measures both worlds.
Exact model version selectable
Not just the engine, but the exact version: gpt-5-high instead of gpt-4o, gemini-3-pro, claude-opus-4-7, deepseek-v4-pro - steerable by audience, industry and intent. So you measure the models your buyers actually use, and re-test reproducibly thanks to fixed versions. No competitor offers this control.
Steered prompts, powered by aipa
At the core of aiva runs aipa, our reproducible prompt infrastructure: AI-generated but deliberately steered questions per industry, persona, intent and use case - versioned and repeatable identically at any time. One prompt, all models in parallel, clean comparability instead of chance. aipa is an integral part of aiva, not a separate license.
Analysis & audit
Quantitative evaluation of every LLM answer: mention, positioning, sentiment and cited sources - segmented by brand and non-brand. Plus a multi-layered brand and competitor logic: parent company, subsidiaries, product brands and generic material terms in parallel, with dedicated competitor sets per industry, country and language. No domain-centric SEO grid.
Open methodology & before/after steering
Dimensions, personas and prompts are laid out openly as Excel - reproducible and auditable. The model is a controlled before/after test: baseline, targeted measures, re-test at identical configuration and KPIs per dimension. Visibility becomes steerable, not just observable.
Recommendations included
Prioritized recommendations with a business case per measure, plus a workshop. The strategic interpretation comes along as consulting - not as a signal you have to interpret yourself.
A typical finding from the aiva analysis
- Brand score: 6.5 / 10 - strong recognition on branded searches
- Non-brand score: 1.5 / 10 - critical gap on generic searches
- Score gap to the competition: +3.4 (distance to the leading provider)
- Non-brand share: 99 % of all search intents without a brand mention
Included in every aiva analysis
- Multi-LLM measurement with aiva - Western (ChatGPT, Perplexity, Gemini, Claude, Copilot) and Chinese (DeepSeek, Doubao, Ernie, Qwen, Hunyuan, Kimi)
- Brand vs. non-brand segmentation
- Opportunity map by impact × effort
- Three fields of action: technical, content, third-party sources
- Coaching, workshops and execution - also with your agencies
- Impact control: re-measurement and tracking after 3 and 6 months