evidoria

← Back to browse

Good practice Imported

Offline On-Device AI Tutor Reaches Indigenous-Language and At-Risk Students at Esperanza Juvenil, Guatemala City

Guatemala · Guatemala City · See the Guatemala profile · See the Guatemala City profile

Evidence: Descriptive / self-reported Top 89% 27/100 · Ask Evidence Copilot about this practice

Esperanza Juvenil, a residential school for 170 at-risk children in Guatemala City (a third with Mayan first languages), deployed NPU laptops in 2025 running a fully offline AI tutor for Spanish, English and maths, donated by Intel and World Wide Technology.

170
Students served at Esperanza Juvenil
~33 % (roughly a third)
Students with a Mayan language as their first language
57 %
Guatemala City metro-area residents below the poverty line
3.2 million
Guatemala City metropolitan-area population
Offline On-Device AI Tutor Reaches Indigenous-Language and At-Risk Students at Esperanza Juvenil, Guatemala City

Details

Maturity
Pilot
Promoter
Esperanza Juvenil (Boys Hope Girls Hope Guatemala), with Intel and World Wide Technology
Period
July 2025–ongoing (early pilot)
Keywords
nonprofit education, indigenous-language inclusion, offline ed-tech, poverty alleviation

Context

Esperanza Juvenil, a residential school in Guatemala City affiliated with Boys Hope Girls Hope, serves 170 academically capable children from severely at-risk backgrounds, roughly a third of whom speak a Mayan language (Mam, Tz'utujil, K'iche' or Kaqchikel) as their first language rather than Spanish. Intel notes that 57% of the roughly 3.2 million residents of the Guatemala City metropolitan area live below the poverty line, framing the offline AI tutor as a way to reach students where reliable internet access cannot be assumed.

Objectives

Support students' Spanish, English and mathematics learning via a fully offline AI tutor that does not depend on internet connectivity.

Activities

In 2025, Intel and World Wide Technology donated laptops built around neural processing units capable of running a large-language-model tutor entirely offline. Teachers are reported to review the AI tutor's chat logs to adjust lesson plans to individual students' needs.

Conclusions

No participation, usage or learning-outcome data has been published for the tutor itself — only a description of the rollout and quotes about its intended purpose. The deployment remains confined to Esperanza Juvenil's single 170-student campus with no announced plan for wider replication.

Implementation
Indicative cost
Low (< €50k) — Laptops with onboard NPUs donated by Intel and World Wide Technology; no monetary donation value disclosed.
Time to results
Short (< 1 year) — Deployed July 2025; as of the sources reviewed (about two months post-launch) no outcome data had been published.
Staffing & skills
Esperanza Juvenil teaching staff (review AI chat logs to personalise lesson plans), Intel and World Wide Technology (hardware donation and deployment partners)

Conditions for success

  • NPU-equipped laptops enabling fully offline, on-device LLM operation without internet dependency
  • Donated hardware from Intel and World Wide Technology removing the upfront cost barrier for the school

Common failure modes

  • No training programme, curriculum, or capability-building framework for teachers has been documented
  • No announced plan for wider replication beyond the single 170-student campus

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