evidoria

← Back to browse

Good practice Imported

Bookbot — Offline AI Reading Companion for Children with Dyslexia in Indonesia

Indonesia · Jakarta · See the Indonesia profile · See the Jakarta profile

Evidence: Descriptive / self-reported Top 24% 60/100 · Ask Evidence Copilot about this practice

Bookbot is an offline AI reading app giving Indonesian children real-time pronunciation feedback on-device; built with UNICEF Venture Fund support and recognised as a UN Digital Public Good, it targets 2.5 million primary pupils with dyslexia and other learning difficulties.

29.4 million
Indonesian primary-school children targeted
2.5 million
Estimated children with learning difficulties (dyslexia, visual impairment) targeted
Bookbot — Offline AI Reading Companion for Children with Dyslexia in Indonesia

Details

Maturity
Pilot
Promoter
Bookbot (with UNICEF Venture Fund, INOVASI)
Period
2018–present (Indonesia rollout from 2021)
Keywords
literacy, assistive technology, speech recognition, education technology

Context

Indonesia has 29.4 million primary-school children, including an estimated 2.5 million with learning difficulties such as dyslexia and visual impairment, and many rural schools have low or no internet connectivity.

Objectives

Give children real-time pronunciation feedback while reading aloud, in a form that works without an internet connection.

Activities

Bookbot is an offline, on-device AI reading-companion app that listens to children read aloud, analyses pronunciation in real time, and gives instant feedback. Developed with support from the UNICEF Venture Fund and Australia's INOVASI programme, Bookbot built a dedicated Indonesian Phoneme Recognizer for Bahasa Indonesia, recognised as a UN Digital Public Good, and formalised localisation through a 2021 government-linked contract.

Results

UNICEF and MIT Solve profiles describe children with dyslexia and visual impairments progressing through the app's book-levelling system, and teachers in rural Sumenep schools reporting engagement gains.

Conclusions

No independently verified, peer-reviewed impact study was found; a percentage improvement figure circulated in the funder's own communications could not be independently corroborated and is therefore not cited here.

Implementation
Indicative cost
Low (< €50k) — Seed-funded through the UNICEF Venture Fund with support from Australia's INOVASI programme; on-device architecture avoids ongoing cloud/connectivity costs.
Time to results
Medium (1–3 years) — In development since 2018; Indonesia rollout and government-linked localisation contract from 2021.
Staffing & skills
Bookbot development team, UNICEF Venture Fund as seed funder, Australia's INOVASI programme as co-funder/partner, 2021 government-linked contract partner for localisation

Conditions for success

  • Offline, on-device processing suited to Indonesia's low-connectivity rural schools
  • Dedicated Indonesian Phoneme Recognizer built for Bahasa Indonesia rather than adapting a generic model
  • Released as a UN Digital Public Good, supporting reuse and scrutiny

Common failure modes

  • No independently verified, peer-reviewed impact study was found; evidence is qualitative case reports from UNICEF/MIT Solve profiles and teacher comments in rural Sumenep schools
  • A percentage improvement figure circulated in the funder's own communications could not be independently corroborated and is not cited

Commonly funded by

Philanthropic / foundation funding

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Attachments

Similar practices you may find useful