From the email
Call me now
The participant opens a secure link from the study's invitation, checks their phone number and taps "Call me now". The AI interviewer calls straight back.
An AI interviewer calls your participants and asks your questionnaire as written. Two models check every answer, and your team reviews anything they don't agree on.
Long running studies move more of their data collection online with every wave. Many participants switch to a web survey or a study app. A large group never will, and in studies of ageing they're often the people you most need to hear from.
For them, computer assisted telephone interviewing (CATI) is the usual answer. It works, but it needs trained interviewers, evening and weekend shifts, call backs and supervision. A 90 minute interview for thousands of people is a large contract, and every interviewer reads the questions a little differently.
WeGuide's AI telephone interviews keep the call and drop the bottleneck. Participants still talk to a voice on the phone. The questionnaire, the checks and the audit trail run on the platform.
Older participants are rightly wary of unexpected calls. So the first contact is an email or letter from the study team they already know, and the participant starts the call.
From the email
The participant opens a secure link from the study's invitation, checks their phone number and taps "Call me now". The AI interviewer calls straight back.
Phone only
The letter or email gives a study phone number and a personal code. The participant rings when it suits them, and the code links the call to their record.
Booked
The participant picks a time. The interviewer calls from a number named in the invitation, and reads out a code that matches the one they were sent.
However the call starts, the interviewer says it's an AI first, offers a person instead, and asks the same questionnaire. The answers get the same checks and land in the same record.
You bring the instrument. We build it as a voice interview in the same Program Builder and Form Builder you use for any WeGuide study.
Every question, answer option, routing rule and probe goes into the study. The platform, not the AI, decides which question comes next, so branching follows your protocol exactly.
Participants get the invitation from your study team, with the call options and what to expect. Consent to the call and the recording is part of the flow, alongside your usual eConsent.
The AI interviewer says it's an AI at the start, reads each question as written, and records the participant's own words. It repeats on request, slows down, and can stop and pick up again another day.
After the call, two separate models map the transcript to your answer options without seeing each other's result. Where they agree, the answer is filed. Where they don't, it goes to your team.
Your team sees each flagged answer next to the transcript and audio clip, and decides. Results land in the study dataset, ready for WeGuide Analytics or export through the Integration Engine.
Follow one answer from the call to your dataset. Two agents score it on their own, and it's only filed when they agree.
An answer from a phone call is only as good as the trail behind it. Each one keeps the record you'd need to defend it.
The full transcript and the participant's verbatim answer, linked to the question it belongs to.
Audio stored with the study, so a reviewer can listen to the moment the answer was given.
Both model results, the prompts and model versions used, and whether they agreed.
Who reviewed a flagged answer, what they chose and why, with a time stamp.
Agreement rates, review rates and call length per wave, so you can report on the method itself.
Phone answers sit next to web and app answers from the same people, labelled by how they were collected.
Each method has its place. Most cohorts will use all three, and keep human interviewers for the measures that need them.
| What your study needs | AI telephone interviews | Human CATI | Web survey only |
|---|---|---|---|
| Reaches people who won't do a web survey | Supported:A normal phone call, landline or mobile. | Supported:A normal phone call with a trained interviewer. | Not supported:Needs a device, internet and confidence online. |
| Available when the participant is | Supported:Any time the study allows, including evenings and weekends. | Partly or depends:Within interviewer shifts and booking slots. | Supported:Any time. |
| Questions asked the same way every time | Supported:Wording and routing come from the platform. | Partly or depends:Depends on training and supervision. | Supported:Fixed text on screen. |
| Help when a question isn't clear | Supported: Repeats the question and asks your approved follow-ups, the same way every call. Hands over to a person when needed. | Supported:Interviewer judgement. | Not supported:Help text only. |
| Audit trail per answer | Supported:Transcript, audio, both model checks and any human decision. | Partly or depends:Keyed answers; recordings depend on the provider. | Partly or depends:Answers and time stamps. |
| Cost as the sample grows | Supported:Mostly call minutes, so the cost per completed interview stays low. | Not supported:Grows with interviewer hours, call backs and supervision. | Supported:Low. |
| Measures that need a person in the room | Partly or depends:Not in the call itself. Memory, attention and reaction time tasks run in the WeGuide app, next to your other instruments. Screening tests such as the MoCA and physical measures stay with your team. | Partly or depends:Some cognitive tests are validated by phone. | Not supported:Self report only. |
SupportedPartly, or depends on setupNot supported
Studies of ageing are the hardest test for a voice interviewer. Hearing loss, fatigue and a healthy distrust of strange callers are normal, and the interview has to work anyway.
Use them alongside the app and web surveys in any WeGuide study, including decentralised clinical trials.
Follow up the participants who don't complete the web questionnaire, wave after wave.
Collect outcome questionnaires by phone after people leave the clinic.
Keep registry data current for people who rarely log in.
Offer a phone route for patient reported outcomes, with the same items as the app.
Interview people in the language they're most comfortable in. WeGuide supports more than 50 languages, from the same translated questionnaire.
Include people who'd never download an app, so the evidence covers everyone.
Questions we hear from research teams planning telephone follow-up.
It's a structured phone interview run by an AI voice instead of a human interviewer. The AI reads your questionnaire as written, listens to the answer and records it. The study platform controls which question comes next, and every answer is checked after the call.
Computer assisted telephone interviewing (CATI) uses a human interviewer reading from a script on screen. AI telephone interviews keep the script and the phone call, and replace the interviewer with an AI voice. You still need a person for some measures, and for anyone who asks for one.
Trust comes from the study team, not the voice. Participants get the invitation from a sender they already know, they start the call themselves, and the interviewer says it's an AI straight away. That's why the pilot starts with a small participant panel before any wider rollout.
Any change of mode can shift answers, whether it's paper to web or face to face to phone. We recommend a pilot that compares AI interviews with your current method for the same people, so you can measure and report any mode effect.
They get one more round with each other's reasoning. If they still don't agree, the answer goes to your team's review queue with the transcript and audio clip. Nothing uncertain is filed without a person deciding.
A single call can run for up to two hours. For long questionnaires we recommend shorter sessions, because the interview can stop and pick up where it left off on another call.
Yes. The AI interviewer can hold the interview in more than 50 languages, including English, Spanish, Mandarin, Arabic, Hindi and Vietnamese. It asks the same translated questions as your app and web surveys, so answers in every language land in one dataset. Read more about multilingual support.
Audio, transcripts and answers are stored with your study data in WeGuide, under the same security and access controls as the rest of the platform.
Committees usually ask for AI disclosure at the start of the call, consent to recording, a way to reach a person, a distress protocol and how safeguarding disclosures are handled. We help you write these into your application.
WeGuide is regulated medical software (TGA Class I) with ISO 27001 certification. It supports HIPAA, GDPR and 21 CFR Part 11 requirements.
Send us your questionnaire. We'll set up a pilot with a small group from your participant panel and share the results with you.