Fortell Impact — Hear from everyone your programme reached

What we do

Two things, over the same conversation.

Whether the goal is to hear from the people your programme reached, or to reach them with training and support — it's the same voice-AI infrastructure underneath.

01 — Collect

Outcome evidence, direct from your clients

Qualitative and quantitative data — in any language, with no internet, no smartphone and no literacy required. Baselines, client feedback, complaints, impact evaluations, national consultations.

Phone-based voice agents — call anyone with a handset
Browser-based avatars — structured surveys, on any screen
Live dashboard as data comes in, report at the end

02 — Deliver

Coaching and support, at real scale

An AI coach that remembers who someone is, what they do, and what they struggled with last time. A conversation, not a recorded message — people ask it questions, and call it back in the evening.

Sector-specific coaching — entrepreneurship, health, protection
Inbound hotlines, outbound calls, SMS — as suits your people
Every interaction also becomes evidence for your board

Why a conversation beats a form. A form gets you the answers to the questions you thought to ask. A conversation surfaces what people raise unprompted — issues nobody thought to ask about, like repayment dates falling before market days. That's the finding that changes a programme, and no questionnaire would ever have caught it.

Why work with us

The advantages, in plain terms.

Voice AI is only useful if the data holds up, the price makes it repeatable, and the ethics are watertight. Here's how we handle each.

01 — Price

€

A fraction of enumerator cost

A fraction of enumerator cost — and it falls with every model release.

02 — Repeatable

4×

A quarterly rhythm, not a one-off

Repeat rounds cost about half — agent, questions and language already exist.

03 — Fast

Days

Days, not months

Days or weeks, while the decision is still open. UNICEF Mozambique: 2,300+ interviews in under a week.

04 — Scale

100k

Hundred, or hundred thousand

100 conversations or 100,000 — same week either way.

05 — Language

200+

In their language, verified

Any language, including dialects. Verified by a native speaker before it goes live.

06 — Ethics

GDPR

Consent, data, and dignity

Informed consent on every call. GDPR-equivalent storage, deletable on request. You never hand over a contact list unless you want to.

How the data holds up. Our coding of these conversations agrees with independent human experts at κ = 0.75, verified with Jhpiego — above the 0.61 threshold for substantial agreement. The data stands up to scrutiny from academic reviewers, boards and funders.

01 · Collecting the data — two demos you can try now

An avatar in the browser. A voice agent on the phone.

Two channels, same underlying AI. Try one and hear the difference from a form in about two minutes.

Demo · Browser avatar

The Avatar

Where in the browser, face-to-face on a screen
Built for structured surveys — closed and open-ended questions
Language your client's own words, in their own language
Rigour strict enough for baselines and impact evaluations; flexible enough to hear what people actually mean

Needs: screen + internetBest for: structured surveys

Speak with one of our avatars →

Demo · Phone voice agent

The Voice Agent

Where over a plain phone call, on any handset
Requires no smartphone, no internet, no literacy
Best for the clients you'd otherwise miss
Rigour choose how rigid the script is — probe like an ethnographer, or stick to a fixed questionnaire

Needs: any handsetBest for: hardest-to-reach

Speak with one of our agents →

Which one? If your clients have a smartphone and a moment of quiet, the avatar is faster to launch — you share a link, they never hand over a contact list. If they're on a basic phone, in the field, or would never click a link, the voice agent calls them.

02 · Delivering the coaching — two demos you can try now

An AI coach that remembers, and calls back.

A conversation the participant can steer — about their business, their health, their case — that picks up where it left off next week. Try it in English, or in Swahili.

Demo · Business coach · English

Coach a microentrepreneur through her week

Built on the same approach as the coaching agent we run with Shortlist and Inkomoko, funded by the Gates Foundation. Ask it how to find your first customer, or how to price a product. It'll remember what you said, and follow up on your commitments.

Under the hood: your curriculum, delivered as a real conversation in the participant's language — with baseline, midline and endline data captured automatically as the coaching runs.

Open the coach →

1:1

Personalised — every participant, every call. The agent knows their business, their family, their last conversation.

24/7

On their schedule — they call it in the evening, when the shop is shut, when the kids are asleep.

Demo · Juma · Swahili

Meet Juma — the same coach, in Swahili

Same architecture, delivered in the participant's own language. Juma coaches microentrepreneurs in Swahili — proof that the coaching agent isn't an English product with translation bolted on, but works natively in the languages our participants actually speak.

Why it matters: most digital coaching stops at English, French, Spanish. Ours doesn't — dialects and small languages included.

Open Juma →

SW

Native Swahili — spoken naturally, not translated. Same for 200+ other languages on request.

↻

Same coach, any language — the memory, the follow-up, the curriculum. Just in the language that fits the participant.

Proof — across the institutions we serve

Three institutions. One instrument.

The same voice-AI infrastructure runs a national youth consultation for a UN agency, a portfolio-wide vulnerability assessment for a development bank, and an EU-aligned impact evaluation for a microfinance NGO.

NGO · UN agency

UNICEF · Mozambique

National youth consultation

2,300+

young people
heard in under a week

Young people across Mozambique were asked what they were struggling with and what UNICEF should do differently — conventionally a months-long exercise, delivered while the decisions were still open.

What it was National consultation. Browser avatar, distributed through UNICEF's existing youth networks; Portuguese and local languages.

Read the case study →

Development bank

IDB Lab · Colombia

Portfolio vulnerability assessment

1 bank

many investees,
one comparable instrument

We interviewed the end-beneficiaries of a microfinance institution in IDB Lab's Colombian portfolio — complementing the investee's own reporting by hearing directly from the women and small-business owners the bank's capital ultimately reaches.

What it was Vulnerability assessment of an MFI's end-clients, on behalf of IDB Lab. Voice agents by phone; Spanish, aligned to IDB Lab's vulnerability framework.

Read the case study →

Microfinance · MFI

Treball Solidari · Spain

Impact evaluation, inside the EU

89%

completion rate,
292 women reached

An end-to-end impact evaluation for Treball Solidari, our client — a microfinance programme for women in vulnerable situations. Spanish voice agents; indicators mapped to the European Pillar of Social Rights. None of them had ever spoken a questionnaire aloud before.

What it was Full impact evaluation for Treball Solidari — baseline, endline, outcome mapping. Phone voice agent; Spanish, EU-aligned indicators.

Read the Treball Solidari case study →

The VoicePrint framework

Three layers that turn a conversation into comparable evidence.

Every VoicePrint interview captures the same three layers — so answers stay comparable across people, rounds, and countries, without flattening what anyone actually said.

Layer 01 · Profile

Who is speaking

Cohort attributes — gender, age, geography, sector, vulnerability markers — so every finding can be disaggregated the way your framework requires.

Layer 02 · Outcomes

What changed

Structured indicators against OECD DAC, IRIS+, EPSR, or your own logframe. Closed answers where you need comparability, scales where you need direction.

Layer 03 · Voice

Why it changed

Open-ended reflections in the respondent's own language and words. Clustered into themes with verbatim quotes attached to every number.

Inside the pipeline

From raw voice to auditable evidence.

Every VoicePrint response moves through the same six stages — so what comes out the other end is structured, traceable, and defensible.

01 · Transcription

Speech to text

Tuned for low-resource languages and regional accents. Timestamped and traceable back to the source audio.

02 · Translation

Into a working language

Translated for analysis, original-language transcript preserved. Verified by a native speaker.

03 · Coding

Against your framework

OECD DAC, IRIS+, EPSR, or your own logframe. AI first pass, human adjudication, inter-rater agreement reported.

04 · Analysis

Indicators & themes

Closed answers aggregated into indicators. Open answers clustered into themes with verbatim quotes, segmented by cohort.

05 · Reporting

Dashboard + report

A live dashboard while it runs, a written report when it's done. Everything exportable — no black box.

06 · Quality assurance

Every stage checked

Each stage spot-checked against the one below. Discrepancies logged and resolved before anything reaches the report.

What you get

A live dashboard while it runs. A report when it's done.

You don't wait months to see what's coming back. Every conversation lands in your dashboard as it happens — your indicators, your themes, and the verbatim voices behind every number.

Deliverable 01 — from day one

Your dashboard

Track your own indicators as the conversations come in — and see them move from round to round, not as a one-off snapshot.

Your indicatorsLive · Round 3

1,842

Conversations
this round

94%

Completion
rate

11

Languages
covered

Income increased since joining68% ▲9

Repayment is not a burden81% ▲3

Understood the loan terms74% ▼4

Your indicators, the themes behind them, and the verbatim voices — updating live as conversations come in.

Deliverable 02 — at the end

Your report

Findings against your indicators, the themes, and the verbatim voices behind every number. Ready to put in front of a board or a funder.

And here's a finished one — download

Two real deliverables, from two different channels.

One collected by a voice agent over the phone, one by an avatar on a screen.

Voice Agent · Phone

GAIL — Kenya

Baseline for our GAIL coaching programme with women microentrepreneurs

Baseline study for GAIL, our coaching programme in Kenya with women microentrepreneurs — part of the M&E that lets us evaluate the impact of the coaching. Collected by a conversational voice agent over the phone; no smartphones or internet on the respondent side.

Download PDF

Avatar · Screen