evidoria

← Back to browse

Good practice Imported

Eligiendo Mi Camino — AI Math Tutor and Career Coach, Lima, Peru

Peru · Lima · See the Peru profile · See the Lima profile

Evidence: Descriptive / self-reported Top 42% 53/100 · Ask Evidence Copilot about this practice

The World Bank and Peru's Ministry of Education, with uDocz, Anthropic and Microsoft, launched an AI math tutor and career-guidance coach reaching 85 public schools and about 4,500 fifth-year students in Lima from March 2026, alongside a 1,270-teacher randomised trial and AI-assi

~4,500 students
Fifth-year secondary students reached by the AI tutor and career coach (since March 2026)
85 schools
Public schools reached in Lima (since March 2026)
1,270 teachers
Teachers in the World Bank randomised controlled trial
6,000+ school leaders
School leaders trained via El Director Libre
13 %
Peruvian secondary students reaching expected math proficiency (baseline)
Eligiendo Mi Camino — AI Math Tutor and Career Coach, Lima, Peru

Details

Maturity
Pilot
Promoter
World Bank Group; Peru Ministry of Education; uDocz; Anthropic; Microsoft
Period
March 2026 – ongoing
Keywords
K-12, tutoring, career guidance, teacher training, public education

Context

Eligiendo Mi Camino ('Choosing My Path') is a joint initiative of the World Bank Group and Peru's Ministry of Education, built with edtech partner uDocz and AI providers Anthropic and Microsoft, against a baseline where only 13% of Peruvian secondary students reach expected math proficiency.

Objectives

The programme combines a personalised AI math tutor with an AI career-guidance coach for secondary students, plus an AI-assisted classroom-observation tool and the 'El Director Libre' methodology, which uses AI to free school principals from paperwork for instructional leadership.

Activities

Since launching in March 2026, the tutor and career coach have reached roughly 4,500 fifth-year secondary students across 85 public schools in Lima, the World Bank is running a randomised controlled trial of practice-oriented teacher training with 1,270 teachers, and El Director Libre has trained more than 6,000 school leaders nationally.

Results

Early results show AI-generated classroom-observation feedback comparable to trained human observers, though outcome data on student math-proficiency gains from the ongoing randomised controlled trial had not yet been published as of this review.

Conclusions

The World Bank's own reporting is candid that technology alone rarely changes outcomes, and that uptake required pairing the AI tools with structured teacher training and institutional support.

Implementation
Indicative cost
Medium (€50k–€500k) — No public budget figure is available; the initiative is jointly resourced by the World Bank, Peru's Ministry of Education and technology partners uDocz, Anthropic and Microsoft.
Time to results
Medium (1–3 years) — The programme launched in March 2026 and was still in early deployment as of this review, with the teacher-training randomised controlled trial and its outcome publication still ongoing.
Staffing & skills
World Bank Group, Peru's Ministry of Education, uDocz (edtech partner), Anthropic and Microsoft (AI providers)

Conditions for success

  • pairing AI tools with structured teacher training and institutional support, since technology alone rarely changes outcomes per the World Bank's own account
  • multi-partner backing combining a development institution, a ministry and AI/edtech providers
  • an existing national school-leader training pipeline (El Director Libre) to build institutional capacity

Common failure modes

  • technology alone does not change outcomes without accompanying teacher training and institutional support, per the World Bank's own reporting
  • student learning-outcome data from the randomised controlled trial had not yet been published as of this review, so measured impact remains unconfirmed

Commonly funded by

National / regional programmes

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