Skip to main content

​
Getting to know voi

voi is an AI impact researcher. You describe your project and impact goals. voi generates a scientific interview guide in minutes. voi conducts interviews with the people you serve, at scale, letting them answer by text, voice, or video. voi turns the results into verified evidence you can act on, share, and be proud of.
voi supports a diverse audience, including:
  • Social enterprises proving their impact without needing consultants or a research team
  • Capital providers verifying impact across a portfolio and strengthening due diligence for LP reporting
  • Accelerators and incubators capturing the real experiences of their cohorts
  • Academics and researchers running rigorous beneficiary-level studies at a scale that was previously too costly
  • NGOs and civil society organisations strengthening monitoring and evaluation with verified stakeholder evidence
  • Policymakers and public authorities turning citizen and community experience into evidence for smarter programme design
A generic survey builder gives you a blank form and leaves the impact methodology to you. As an impact researcher, voi supports the full process. voi builds your Theory of Change. voi recommends indicators from a library based on research. voi drafts the questions. voi runs the interview itself, asking live follow-up questions based on what a respondent just said. Respondents have a conversation, rather than filling out a form. voi turns hundreds of open-ended answers into indicator scores, automatically clustered themes with representative quotes, and a report draft. Every part of this is traceable back to the original response. None of this exists in a standard form builder.
No. You describe your project in plain language, and voi drafts a Theory of Change and suggests indicators and questions for you. You’re free to edit everything if you already know what you want to measure, but the platform works as well for a first-time user as for a trained M&E practitioner.
Yes. A fund can measure impact across its portfolio companies and compare results across the organisations it funds. A dedicated view for comparing several organisations at once is in development.
Today they’re separate products: voi is the AI interview and analysis engine, while leonardo’s Impact Management Platform is a broader tool for measuring, verifying, and reporting impact and ESG performance.

​
Designing your study

You describe your project: who you serve, what you do, and what you are trying to change. voi drafts a Theory of Change from this description. voi then recommends indicators from its library for each outcome, with matching questions. voi uses a library indicator only where it fits an outcome strongly, and writes an AI-generated question for any outcome the library does not cover that well. See Build a Theory of Change and How voi writes your interview.
Yes. Each indicator in voi’s library traces back to where it was originally defined. This may be a published scale, an academic paper, or a framework. It does not trace back to a secondary report that cited the original source. This means you can see exactly which standard your results map to. See How indicators work.

​
Collecting responses

By typing, speaking, or recording a short video, question by question. Closed questions (scales, single or multiple choice, NPS) are answered directly; open-ended questions can be answered in whichever format is most comfortable for the respondent.
Respondents can complete a study in their own language, with open-ended answers translated automatically. See Languages & accessibility for the full list.
Voice and video response modes mean respondents don’t need to type an answer, and for a subset of languages, voi can read the question aloud too. See Languages & accessibility.
The default is generating a link and sending it over a digital channel; respondents complete it on their own device, in their own time. When that’s not realistic, voi supports a person-in-the-loop collection method instead. See Collection methods.
It depends on your question mix, since open-ended answers by voice take longer than tapping a scale. Without a target, voi covers every outcome first and keeps the interview as short as that allows. You can set a target length in Plan, and voi then keeps the interview within it, telling you if an outcome does not fit and how you could make room. See Setting a target completion time.
Not as a built-in feature today. voi’s flow generates and runs the conversation itself end to end; feeding in an independently conducted interview for the same automated theming and scoring isn’t a supported path.
No. voi runs no emotion recognition and no voice-based inference of any kind. Recordings are used only to produce a transcript, and the analysis works from that text. See Voice, video & AI.

​
Turning answers into insight

Every open-text answer is grouped into themes automatically, each with a representative quote, so you see the topics that came up most instead of reading every response individually. You can also disaggregate any result by demographic group or by how someone answered a different question.
Not yet. Longitudinal tracking is one of the most requested features from pilot organisations and is on the roadmap. Today each study is analysed on its own.
Yes. The Explore feature lets you chat with your study’s results directly, in plain language. See Explore.
Yes. voi drafts a report directly from your analysis, and you can customise it: reorder or remove sections, choose which indicators and quotes to feature, and use voi’s AI writer to draft or refine narrative text. See Report.

​
Data & privacy

For anonymity, consent, AI processing, and data retention, see Compliance & security. For which companies process your study’s data and how they’re chosen, see AI models & providers.