Detailed Planet-Project Plan — Africa's EdTech Breakthrough
Every RESPECT Compatible App needs access to the same five capabilities — localization into African languages, adaptive learning, curriculum-aligned assessment, accessibility for learners with disabilities, and evidence-based engagement mechanics — and today every developer builds each one independently, at high cost, or omits it entirely. PREMIER is a seven-year applied research institute that will build each capability once, as shared platform infrastructure, collapsing the per-app cost for all RESPECT Compatible Apps simultaneously. Each PREMIER research project follows the same pattern: identify a capability that is currently expensive, manual, and siloed; build it into the RESPECT Platform as a shared service, API, or library; and validate it in African classroom settings. Individual research projects are staggered across the seven-year period as their prerequisites (CRADLE's federated data, ECM's Curriculum IR) mature.
The PREMIER Institute houses two categories of research projects:
The Little Easies — five research projects bundled with the Institute's founding:
The Big Easies — two major IR-based research projects housed within the Institute but independently fundable:
Founding the PREMIER Institute means funding the Institute itself plus all five Little Easies. The Institute Founder receives naming and hosting rights for the Institute and for any institutions that the Little Easies produce. The Big Easies are funded separately, each with its own Founder attribution: the donor who funds the first tranche of a Big Easy receives Founder attribution for that project and naming/hosting rights for any institution it produces (e.g., ECM produces the SOCLE Board). The Institute Founder has Right of First Refusal on both Big Easies.
Each project produces shared platform infrastructure — APIs, libraries, models, and specifications — that RESPECT Compatible App developers will use through the RESPECT Platform, eliminating the need for each developer to build these capabilities independently. The research institute model ensures that each capability is grounded in rigorous evidence, validated in classroom settings, and designed for the linguistic, cultural, and infrastructural realities of African education systems.
Duration: 84 months (seven years, three phases aligned with the Breakthrough Project's V&P_Core Tranches).
Requested funding: USD 27M (Institute operations plus the five Little Easies). The Big Easies (ECM: USD 10M; Easy FLN Localization: USD 8M) are funded separately.
Natural Development Partners: Research-focused Development Partners — Wellcome Trust (education and learning science), IDRC (Canada), Swedish Research Council (Vetenskapsrådet), Swiss National Science Foundation, NORHED (Norway), and the Gates Foundation (for projects with direct overlap with ECM and learning outcomes measurement).
The Institute Founder funds PREMIER's core operations (USD 3.5M) plus the five Little Easies (USD 23.5M), for a total commitment of USD 27M over 84 months, disbursed in three tranches aligned with the Breakthrough Project's V&P_Core phases.
Phase 1 success is defined by the Month-24 Proof of Capability outcome set (Section 1B). If Phase 1 deliverables are met, Phase 2 funding is released. If they are not met, PREMIER and the funder jointly review scope and timeline before proceeding.
PREMIER reports to the Breakthrough Project's governance structure. The Institute Director reports quarterly to the funder and annually to the Breakthrough Steering Committee. Independent evaluations at Month 36 and Month 78 provide external accountability (see Section 9).
PREMIER Institute owns all intellectual property resulting from Institute-hosted research projects. Funding partners receive a worldwide, paid-up, royalty-free, sub-licensable, non-exclusive license to all such IP. Research outputs — models, datasets, code, and specifications — are published under open-source and open-access licenses. University research partners retain academic publication rights; all code and infrastructure deliverables are owned by PREMIER.
Attribution is distinct from authority. Naming rights, hosting rights, and Legacy Attribution are recognition mechanisms; they confer no governance authority over research agendas, data access, or platform operations. Governance authority remains with the Breakthrough Project's established structures: AUDA-NEPAD for continental coordination, national governments for data sovereignty, and the RESPECT Platform's technical steward for infrastructure decisions. The Institute Founder's preferred host country is prioritized, but governance is unchanged regardless of host location.
The following concrete outcomes define Phase 1 success and gate Phase 2 funding:
Easy Knowledge Assessment (launching Year 3) and Easy Courseware Gamification (launching Year 4) do not have Phase 1 deliverables; their first milestone gates appear in the Go/No-Go Gates table (Section 10.2).
Africa's EdTech ecosystem suffers from a structural R&D problem. Capabilities that every learning app needs — localization into African languages, adaptive pathways, valid assessment, accessibility for learners with disabilities, and evidence-based engagement mechanics — are currently developed independently by each developer, for each app, in each language. This duplication is expensive, inconsistent, and often results in these capabilities being omitted entirely from products serving African learners.
The pattern is visible across multiple dimensions. Localization is the most acute: Africa's 2,000+ languages and hundreds of national curricula create an N×M combinatorial challenge that no individual developer can address at scale. Adaptive learning algorithms require large datasets to train effectively, but each developer's user base is too small to generate them. Assessment generation requires curriculum alignment expertise that most developers lack. Accessibility is routinely deprioritized because the per-app cost of implementation is disproportionate to the perceived market size. Gamification is implemented ad hoc, often without evidence of effectiveness and sometimes with harmful engagement patterns.
PREMIER will resolve this structural duplication by building each capability once, as shared infrastructure within the RESPECT Platform, and making it available to all RESPECT Compatible Apps through standardized APIs and libraries. The ECM model — demonstrated in the ECM Research Proposal for curriculum mapping — establishes the pattern: identify a capability that is currently expensive, manual, and siloed; research the best possible implementation; build it into the platform; and validate that it achieves its objectives.
PREMIER's research projects are applied, not open-ended. Each project begins with a defined problem, proceeds through design, prototyping, and validation, and concludes by delivering working infrastructure that is integrated into the RESPECT Platform and available to every RESPECT Compatible App developer. Research questions are resolved through design, prototyping, and validation within defined timescales.
The cost-collapse mechanism operates through three integration modes, depending on the capability:
In all three modes, the economic logic is the same: PREMIER converts a per-app, per-language, per-developer cost into a one-time platform investment that every RESPECT Compatible App inherits upon integration.
Four of the five PREMIER projects will draw on the federated learning data infrastructure established by CRADLE (see the CRADLE Research Program proposal). Easy Personalized Learning requires large-scale, cross-jurisdictional learner interaction data to train and validate adaptive models. Easy Knowledge Assessment requires diverse assessment response data to validate item quality and detect bias. Easy Text Localization requires usage data across linguistic contexts to evaluate translation quality and learner comprehension. Easy Accessibility requires interaction data from learners with disabilities across diverse contexts to validate infrastructure effectiveness.
CRADLE's federated data architecture — nationally sovereign, Malabo Convention–compliant, and accessible to researchers through tiered governance — provides the data foundation that makes PREMIER's research projects tractable at continental scale. Without federated data, each project would require bespoke data collection arrangements with individual countries, at prohibitive cost and delay.
PREMIER will operate as a coordinating research institute, not as a large centralized laboratory. The institute will maintain a small core team (director, research coordinator, administrative staff, and shared technical infrastructure) and will execute research projects primarily through partnerships with African and international universities and research institutions. This structure maximizes research quality, builds African research capacity, and minimizes overhead.
The core team will be responsible for research portfolio management, quality assurance, platform integration coordination, and stakeholder engagement. Research execution — data collection, model development, validation studies, and prototype engineering — will be conducted by partner institutions under subgrant agreements, supervised by project-specific principal investigators.
Core team (6–10 FTE across seven years): Institute Director, Research Coordinator, Platform Integration Lead, 2–3 Project Managers (scaling with active projects), and administrative/finance staff. Core team costs are covered by the Institute core operations budget (USD 3.5M over 84 months).
Institute Director (1.0 FTE, Years 1–7): The Institute Director is PREMIER's chief executive and principal scientific officer. The Director reports quarterly to the funding partner(s) and annually to the Breakthrough Steering Committee, and is recruited during Phase 1 (Months 1–6).
Core responsibilities:
Qualifications profile: The Director will hold a doctoral degree (or equivalent professional experience) in a relevant field (computer science, computational linguistics, learning science, educational technology, or a related discipline). The role requires a demonstrated track record of leading multi-institution applied research programs with budgets of USD 5M or more, experience translating research outputs into deployable software or infrastructure, familiarity with the African higher education and research landscape, and experience managing donor-funded programs with milestone-based disbursement.
Comparable roles at peer institutions include: the Executive Director of CMU's Simon Initiative (who leads a 14-member research and learning engineering team and coordinates applied research across multiple university departments); the Director of Research and Programme Delivery at ESSA (who manages a distributed research team across seven African and European countries and oversees donor-funded projects); and directors of IDRC-funded research hubs (who coordinate multi-country applied research networks with university partners). PREMIER's Director will operate at a similar scope — coordinating a distributed research network, managing a USD 27M budget over seven years, and ensuring that research outputs become operational platform infrastructure.
Budget allocation model: Approximately 63% ± 3% of each research project's budget flows to university partner subgrants (researcher salaries, data collection, compute, and fieldwork). The remaining 35–40% covers core team supervision, platform integration engineering, shared infrastructure, and independent evaluation.
Partner selection: University research partners are selected through open competitive calls, evaluated on: demonstrated research capacity in the relevant domain, presence in or partnership with African institutions, prior experience with applied (deployment-oriented) research, and ability to meet PREMIER's open-source and open-access requirements.
Principal Investigator responsibilities: Each research project has a named PI drawn from a partner institution. The PI is responsible for research design, execution quality, deliverable acceptance criteria, and publication. The PI reports to the Institute Director on milestones and to PREMIER's Research Coordinator on platform integration readiness. On platform integration decisions — API specifications, infrastructure standards, and shared-service architecture — the Institute Director holds final authority, informed by the PI's domain expertise. This applies equally to Little Easy PIs and to Big Easy PIs whose projects are housed within the Institute.
Quality assurance: Each project undergoes annual internal technical review (led by the Institute Director and Research Coordinator) and two independent external evaluations (Month 36 and Month 78). Subgrant agreements include deliverable acceptance criteria, milestone-based disbursement, and financial audit provisions.
Easy Text Localization (Years 1–5, USD 3.5M)
This project will build shared AI-assisted localization infrastructure within the RESPECT Platform, enabling rapid, cost-effective translation and adaptation of RESPECT Compatible Apps across African languages. The project extends RESPECT's existing Kotlin Multiplatform localization support by adding machine translation models for low-resource African languages, automated quality estimation, community-driven post-editing workflows, and culturally adaptive content transformation.
Key research questions include: What combination of neural machine translation, transfer learning, and community post-editing produces acceptable localization quality for educational content in languages with fewer than 100,000 parallel sentences? How should localization quality be assessed when native-speaker reviewers are scarce? What cultural adaptation (beyond translation) is required for educational content to be pedagogically effective across diverse African contexts?
The project will partner with the Masakhane NLP research community and African university NLP labs. It will build on existing open models (Meta's NLLB, Google's multilingual models) and contribute training data and fine-tuned models back to the open research community.
Easy Personalized Learning (Years 2–6, USD 6M)
This project will build shared adaptive learning infrastructure within the RESPECT Platform, enabling all RESPECT Compatible Apps to personalize learner pathways using a common, privacy-preserving learner model. Rather than requiring each app developer to build their own adaptive engine — a capability that historically costs USD 75M ± 25M for proprietary systems — the project will create a shared platform service that any RESPECT Compatible App can invoke.
Key research questions include: What learner model representation enables meaningful personalization across diverse app types (literacy, numeracy, science, vocational) while preserving privacy? How can adaptive algorithms be trained effectively using federated data that never leaves national jurisdictions? What is the minimum data volume per learner required for personalization to outperform non-adaptive delivery? How should adaptive recommendations account for intermittent connectivity and offline-first usage patterns?
The project's federated training architecture reflects DPI-Ed's commitment to keeping learning data sovereign while enabling AI development — adaptive models will be trained on African data under Malabo Convention governance, producing sovereign AI capabilities tuned to African learners and curricula (see Essay 12): AI in Africa's DPI-Ed.
The project will draw on CRADLE's federated data infrastructure for model training and validation. It will partner with universities with established adaptive learning research programs and with the broader learning analytics research community.
Easy Knowledge Assessment (Years 3–7, USD 5M)
This project will build shared assessment-generation infrastructure within the RESPECT Platform, enabling automated, curriculum-aligned knowledge assessment across all RESPECT Compatible Apps. The project depends on ECM's Curriculum Intermediate Representation (Curriculum IR) being available, which provides the structured curriculum data needed to generate valid, curriculum-aligned assessment items.
Key research questions include: What item-generation techniques produce valid, reliable assessment items from structured curriculum representations? How can generated items be automatically calibrated for difficulty and discrimination without large-scale field testing? How should assessment infrastructure integrate with GEOS's evidence pipeline to produce finance-grade learning outcome signals? What safeguards prevent gaming or teaching-to-the-test when assessment generation is automated?
The project will partner with psychometricians and assessment researchers at African and international universities. It will integrate with the GEOS Organization's audit framework to ensure that automatically generated assessments meet finance-grade evidence standards.
Easy Accessibility (Years 2–5, USD 3M)
This project will build shared accessibility infrastructure within the RESPECT Platform, enabling all RESPECT Compatible Apps to serve learners with disabilities through platform-level services. Currently, accessibility in African EdTech is nearly non-existent: most apps provide no support for visual, hearing, motor, or cognitive disabilities, and international accessibility standards (WCAG) assume infrastructure and language support that is unavailable in most African contexts.
Key research questions include: What text-to-speech and screen reader architectures are effective for African languages with limited digital resources? How should accessibility services account for the low-bandwidth, offline-first delivery model? What platform-level APIs enable app developers to make their products accessible without specialized expertise? How should accessibility be assessed and certified within the RESPECT Compatible certification framework?
The project will partner with the DAISY Consortium, African disability-rights organizations, and universities with assistive technology research programs.
Easy Courseware Gamification (Years 4–6, USD 2.5M)
This project will develop a research-validated, culturally localizable gamification framework providing shared engagement mechanics across all RESPECT Compatible Apps. The project will specifically avoid persuasive design patterns that exploit psychological vulnerabilities, instead developing engagement mechanics that are evidence-based, age-appropriate, culturally sensitive, and aligned with educational objectives.
Key research questions include: What gamification mechanics improve learning outcomes (not just engagement metrics) in low-resource educational settings? How should gamification be culturally adapted across Africa's diverse contexts without reinforcing stereotypes or inappropriate competitive dynamics? What safeguards prevent gamification from devolving into addictive design patterns? How should the effectiveness of gamification mechanics be measured within the RESPECT Platform's evidence infrastructure?
The project will partner with game-based learning researchers and with African educational psychologists who can validate cultural appropriateness across diverse contexts.
PREMIER depends on and contributes to multiple system components:
Some PREMIER research projects may produce governance or standards bodies that outlive the research project itself:
SOCLE Board (produced by ECM): An expert body that maintains and evolves the Structured Ontology for Curriculum and Learning Expectations — the machine-readable curriculum standard that ECM creates. The SOCLE Board's authority derives from the African Union's endorsement of the curriculum standard; it operates as a technical standards body analogous to W3C working groups. It does not set curriculum policy — that authority remains with national governments and the African Union. The SOCLE Board maintains the standard's technical integrity and manages its evolution.
Accessibility certification requirements (produced by Easy Accessibility): A set of testable accessibility criteria integrated into the RESPECT Compatible certification framework. These criteria are maintained by the RESPECT Platform's technical steward, not by a separate body. National adoption decisions remain with national governments.
PREMIER's role is to produce the research and infrastructure that these institutions require; it does not govern them after handoff. Governance authority flows from the Breakthrough System's established structures (AUDA-NEPAD, national governments, technical stewards), not from the research institute.
All PREMIER research outputs will conform to or align with relevant standards. The distinction matters: "comply with" means PREMIER will implement the standard and test/certify against it; "align to" means PREMIER will follow the standard's design principles and interoperate with its interfaces, adapting where the standard does not fully address African educational contexts.
PREMIER occupies a unique position: it is neither a standalone research program nor a technology development project, but a research institute embedded within a functioning DPI-Ed ecosystem. This positioning provides three advantages that standalone research programs lack:
First, PREMIER's research is immediately deployable. Each project's outputs are integrated directly into the RESPECT Platform, reaching every RESPECT Compatible App and every learner in the RESPECT Ecosystem. The gap between "research finding" and "deployed capability" — which can be years or decades in standalone research — is compressed to months.
Second, PREMIER's research benefits from scale data that no standalone research program can access. CRADLE's federated database, built from actual learner interactions across multiple countries, provides training and validation data at a scale that individual research projects cannot generate.
Third, PREMIER's research agenda is governed by platform need, not by academic fashion. The five research projects were identified because they represent the highest-value shared capabilities for RESPECT Compatible Apps — the capabilities whose absence most constrains the quality, reach, and equity of digital education across Africa.
EdTech Hub (FCDO-funded, £20M over 8 years) provides the closest structural precedent: a multi-year research institute focused on EdTech in low-income countries, operating through university partnerships. EdTech Hub produces evidence syntheses, policy briefs, and research publications intended to inform decision-making. PREMIER extends this model in three respects: its outputs are deployable platform infrastructure (APIs, libraries, models) integrated directly into the RESPECT Platform; its research agenda is governed by platform engineering needs; and its outputs reach learners through that direct integration, compressing the research-to-impact pathway from years to months. The two are complementary — EdTech Hub's evidence can inform PREMIER's research priorities, while PREMIER's platform provides EdTech Hub with a deployment channel for validated interventions.
The Simon Initiative at Carnegie Mellon University (over $100M from multiple funders over decades) demonstrates the long-term value of applied learning science research that produces deployable educational technology. The Simon Initiative operates within a single university, serving CMU's own learning platforms. PREMIER operates across multiple university partners and serves an entire continental ecosystem. PREMIER's total budget (USD 27M for the founding portfolio) is a fraction of the Simon Initiative's cumulative investment, but the deployment multiplier is larger: every PREMIER output reaches every RESPECT Compatible App, across every participating country.
Masakhane (African NLP research community, $1M+ in coordinated grants) demonstrates that distributed, community-driven research can produce high-quality NLP capabilities for African languages. Masakhane operates as a volunteer research network; PREMIER provides funded infrastructure, platform integration, and deployment pathways that Masakhane's community model cannot sustain independently. PREMIER's Easy Text Localization project will partner directly with Masakhane, providing resources and a deployment channel for Masakhane's research outputs.
Research-focused bilateral and multilateral funders are the natural constituency for PREMIER. The institute's applied research model — producing deployable platform infrastructure, not just publications — aligns with Development Partners that prioritize research-to-practice pathways.
PREMIER requires an Institute Director with dual credentials: research leadership (track record of leading multi-institution applied research programs with budgets of USD 5M or more) and platform engineering sensibility (understanding of how research outputs become deployable infrastructure). The Director will hold a doctoral degree or equivalent professional experience in a relevant field, will have familiarity with the African higher education and research landscape, and will have experience managing donor-funded programs with milestone-based disbursement. The Director will be recruited during Months 1–6 of Phase 1 and will serve as both the institute's chief executive and its principal scientific officer. The Director will be supported by the Research Coordinator, Platform Integration Lead, and project-specific principal investigators drawn from partner institutions. See Section 3.1 (Institute Director) for the full role description, responsibilities, and comparable-institution benchmarks.
| Partner | Role | Contribution |
|---|---|---|
| African university NLP labs (via Masakhane network) | Research partner | Low-resource language MT research; training data; model development |
| [African university — learning science] | Research partner | Adaptive learning research; classroom validation; learner model design |
| [African university — assessment research] | Research partner | Item generation research; psychometric validation; assessment quality |
| DAISY Consortium | Technical partner | Accessibility standards expertise; assistive technology architecture |
| [African disability-rights organization(s)] | Validation partner | User testing with learners with disabilities; cultural appropriateness |
| [International university — AI/ML] | Research partner | Federated learning; privacy-preserving ML; adaptive algorithm research |
| RESPECT Platform engineering team | Integration partner | API design; platform integration; shared infrastructure deployment |
| CRADLE | Data partner | Federated data access for training, validation, and research |
| GEOS Organization | Standards partner | Assessment quality standards; evidence pipeline integration |
| ECM project team | Dependency partner | Curriculum IR for Easy Knowledge Assessment |
| Category | USD |
|---|---|
| Easy Text Localization (Years 1–5) | 3,500,000 |
| Easy Personalized Learning (Years 2–6) | 6,000,000 |
| Easy Knowledge Assessment (Years 3–7) | 5,000,000 |
| Easy Accessibility (Years 2–5) | 3,000,000 |
| Easy Courseware Gamification (Years 4–6) | 2,500,000 |
| Institute core operations (Years 1–7) | 3,500,000 |
| Shared compute and research infrastructure (Years 1–7) | 1,000,000 |
| Independent evaluation (2 evaluations) | 300,000 |
| Contingency (~8%) | 2,200,000 |
| Total | 27,000,000 |
| Phase | Period | Amount | Focus |
|---|---|---|---|
| Phase 1 | Months 1–24 | $6,000,000 | Institute establishment; Easy Text Localization launch; Easy Accessibility launch |
| Phase 2 | Months 25–48 | $11,000,000 | Easy Personalized Learning main phase; Easy Knowledge Assessment launch; Easy Courseware Gamification launch |
| Phase 3 | Months 49–84 | $10,000,000 | Validation across all projects; GEOS integration; platform deployment; transition |
| Total | 84 months | $27,000,000 |
Each research project budget includes personnel (principal investigators, researchers, engineers), university partner subgrants, data collection and annotation, compute infrastructure, classroom trials, and travel. The institute core operations budget covers the director, research coordinator, administrative staff, convening program, and stakeholder engagement.
The budget assumes that CRADLE's federated data infrastructure is available from Year 2 and that ECM's Curriculum IR is available from Year 3. Delays in either dependency will shift the corresponding PREMIER project timelines, but total costs are not expected to change materially because research capacity can be redirected to earlier-phase projects.
Contingency (8%) reflects the applied nature of the research: problems are defined, but implementation pathways involve genuine uncertainty. Contingency covers scope adjustments, additional validation rounds, and unforeseen technical challenges.
The budget and timeline rest on the following assumptions. If an assumption proves false, the corresponding bound applies.
| Assumption | Bound (what PREMIER is not promising) |
|---|---|
| CRADLE's federated data infrastructure is operational by Year 2 | If delayed, PREMIER operates in degraded mode (see Section 10.1). Total cost is stable; timelines for data-dependent projects shift. |
| ECM's Curriculum IR is available by Year 3 | If delayed, Easy Knowledge Assessment defers item generation; foundational research proceeds. |
| At least 10 African languages have sufficient parallel text corpora to train initial MT models | Easy Text Localization targets 10 languages in Phase 1, expanding to 25+ in Phase 2. PREMIER does not promise coverage of all 2,000+ African languages; it builds infrastructure that scales as language resources grow. |
| Minimum viable personalization requires ~10,000 learner-hours of interaction data per target population | If data volumes are lower, adaptive models operate in population-level mode (less individual precision, still superior to non-adaptive delivery). |
| University partner institutions can recruit and retain qualified researchers within project budgets | If recruitment proves difficult, PREMIER pairs African institutions with international co-PIs and provides competitive compensation supplements within project budgets. |
| PREMIER does not promise that all five Little Easies will reach full production deployment within 84 months | Each project has defined Phase 1 deliverables (Section 1B). Projects that encounter fundamental feasibility barriers are documented, their partial outputs are published, and resources are redirected to higher-performing projects. |
Two independent evaluations will be conducted:
Evaluations will be conducted by independent evaluators with expertise in education technology research, platform development, and African education systems.
| Risk | Likelihood | Impact | Mitigation |
|---|---|---|---|
| CRADLE federated data delayed — reduces data available for training and validation | Medium | High | Easy Text Localization and Easy Accessibility have lower data dependency and can proceed independently. Easy Personalized Learning can use synthetic and pilot-scale data for initial development, shifting to federated data when available. |
| ECM Curriculum IR delayed — blocks Easy Knowledge Assessment | Medium | Medium | Assessment research can proceed on foundational questions (item generation techniques, validity frameworks) while awaiting Curriculum IR. Timeline shifts but total cost is stable. |
| Low-resource language data scarcity — insufficient parallel text for MT models | Medium | Medium | Partner with Masakhane community for data collection. Use transfer learning from related higher-resource languages. Prioritize languages with existing digital corpora for early wins. |
| Adaptive learning cold-start — insufficient learner data per individual for meaningful personalization | Medium | Medium | Design hybrid models that combine population-level patterns with individual data. Accept graceful degradation in early deployment. |
| University partner capacity — African university research infrastructure may limit execution speed | Medium | Low | Budget includes compute and infrastructure support for partner institutions. Pair African partners with international co-PIs where capacity gaps exist. |
| Researcher recruitment — difficulty attracting world-class AI/ML researchers to education research | Low | High | Competitive compensation within research project budgets. Leverage the unique dataset access (CRADLE) and deployment pathway (RESPECT Platform) as recruitment differentiators. |
| Platform integration complexity — research outputs may be difficult to integrate into production platform | Low | Medium | RESPECT Platform engineering team participates in research design from the outset. API specifications defined before implementation begins. |
PREMIER's two principal external dependencies are CRADLE (federated data) and ECM (Curriculum IR). If either is delayed, PREMIER does not halt — it operates in degraded mode, with reduced scope until the dependency is available.
If CRADLE is delayed:
If ECM's Curriculum IR is delayed:
| Gate | Timing | Condition | Action if Not Met |
|---|---|---|---|
| Phase 2 release | Month 24 | Month-24 Proof of Capability outcomes met (Section 1B) | Joint funder-PREMIER review; scope and timeline adjustment before Phase 2 proceeds |
| Easy Personalized Learning: federated training | Year 2+ | CRADLE federated data infrastructure operational and accessible | Defer federated training; continue with synthetic/pilot data; reassess at next annual review |
| Easy Knowledge Assessment: item generation | Year 3+ | ECM Curriculum IR available and validated | Defer item generation; continue foundational psychometric research; reassess at next annual review |
| Easy Courseware Gamification: effectiveness validation | Year 5+ | CRADLE data and GEOS outcome signals available | Defer large-scale validation; publish framework with limited-scale evidence; reassess at final evaluation |
| Phase 3 release | Month 48 | Mid-term independent evaluation (Month 36) findings addressed; Phase 2 deliverables substantially met | Joint funder-PREMIER review; scope adjustment before Phase 3 proceeds |
PREMIER's sustainability model differs from the other Planet-Projects because its outputs are infrastructure, not ongoing operations. Each research project concludes by delivering working platform infrastructure that is maintained thereafter by the RESPECT Platform engineering team (funded through V&P_Core's trademark and certification revenue). Research outputs are open-source and open-access, ensuring that the global community can build on them independently.
The initial five "Easy X" research projects constitute the founding portfolio. They are funded for seven years and will deliver their infrastructure outputs within that period. The institute's mandate, however, is open-ended — because the universe of hard problems that can be solved through applied R&D, followed by deployment at continental scale on Africa's DPI-Ed, will grow as the platform matures. Each year of operation will generate new data, new usage patterns, new pedagogical questions, and new technological capabilities that create research opportunities that did not previously exist.
The university partnerships and African research capacity built during PREMIER's founding portfolio will persist and deepen, contributing to Africa's long-term research ecosystem in education technology. Future "Easy X" research projects — identified as new questions become tractable — will be funded and executed within the institute, extending its portfolio and its impact.
The five founding projects address the most urgent shared capabilities for RESPECT Compatible Apps today. The research frontier, however, will expand as Africa's DPI-Ed matures. Development Partners routinely publish research agendas identifying the evidence gaps that constrain their education investments. Many of these questions have remained unanswered because answering them requires standardized, cross-country data at a scale and consistency that fragmented education systems have never produced. Africa's DPI-Ed — through CRADLE's federated data, GEOS's standardized outcome signals, ECM's structured curriculum representations, and the RESPECT Platform's uniform data layer — transforms these questions from intractable to tractable.
The following table samples research priorities from today's leading Development Partners — questions that the existence of Africa's DPI-Ed is likely to address, in whole or in part, through PREMIER and its successors.
| Research Question (from DP Agendas) | Source DP(s) | DPI-Ed Infrastructure That Makes It Tractable |
|---|---|---|
| How can systems close the numeracy evidence gap, given that literacy has received disproportionate research investment? | Gates Foundation | GEOS provides standardized, finance-grade numeracy outcome signals across jurisdictions. CRADLE federates cross-country numeracy data, creating the largest standardized dataset of foundational numeracy performance in Africa. |
| What pedagogical practices most effectively improve student outcomes in low-resource settings? | World Bank, FCDO | Standardized xAPI learning interaction data from the RESPECT Platform, linked to GEOS outcome signals through CRADLE, enables large-scale comparative analysis of instructional approaches across countries, curricula, and contexts. |
| How can teacher professional development improve quality, support teacher agency, and sustain impact at scale? | IDRC (KIX), World Bank, FCDO | PROMISE-trained teachers generate measurable before/after data through GEOS outcome signals. CRADLE enables cross-country comparison of teacher training modalities, dosages, and sustained effects. |
| Will this EdTech use lead to sustained impact on learning outcomes — and how can that impact be measured continuously? | EdTech Hub (FCDO/Gates) | GEOS provides continuous, standardized outcome measurement through longitudinal evidence pipelines. The "impact per hour" metric enables comparison across products, countries, and populations. |
| How can EdTech be scaled cost-effectively in low-income contexts? | EdTech Hub, UNESCO | The RESPECT Platform's shared infrastructure (localization, assessment, accessibility, adaptive learning) collapses per-app development and deployment costs. Ecosystem-wide cost data, combined with GEOS outcome signals, enables cost-effectiveness analysis at scale. |
| Does EdTech work for the most marginalized children, and does it enhance equity? | EdTech Hub, GPE, IDRC | CRADLE's federated data can be disaggregated by demographics, geography, disability status, and language — enabling equity analysis across populations that have historically been invisible in education data. |
| How can education data systems be strengthened and data better used for decision-making? | IDRC (KIX), World Bank | CRADLE's federated architecture, GEOS's standardized signals, and EMIS interoperability (via BEINGS) collectively build a continent-scale data system designed for policy use — providing a live, continent-scale reference implementation that policymakers can observe, evaluate, and adapt. |
| What is the added value of digital technology in education — and where are the evidence gaps? | UNESCO (GEM Report 2023) | Multi-country deployment of the same platform (RESPECT) across diverse contexts enables controlled comparison of digital vs. non-digital instruction using standardized outcome measures (GEOS), at a scale that no single-country study can achieve. |
| How can adaptive learning algorithms be trained effectively in low-resource, multilingual, offline-first contexts? | IDRC, Jacobs Foundation | Easy Personalized Learning addresses this directly, using CRADLE's federated data for model training without extracting data from national jurisdictions. The offline-first constraint and African language diversity create research contributions with global applicability. |
| What item-generation techniques produce valid, reliable assessments from structured curriculum representations? | Gates Foundation, World Bank | Easy Knowledge Assessment uses ECM's Curriculum Intermediate Representation — the first continent-scale, machine-readable curriculum standard — to generate and validate assessment items across curricula, languages, and grade levels. |
| What gamification mechanics improve learning outcomes (not just engagement) in diverse cultural contexts? | LEGO Foundation | Easy Courseware Gamification tests engagement mechanics against GEOS outcome signals across Africa's cultural diversity, producing evidence on what works, for whom, and in what context — at a scale unavailable to any single-country study. |
| How do children's individual learning needs differ within group settings, and how can education systems address this variability? | Jacobs Foundation (LEVANTE) | Easy Personalized Learning's shared learner model, trained on CRADLE's federated data, generates continent-scale evidence on within-group and within-person learning variability — directly contributing to the Jacobs Foundation's Learning Variability research program. |
| What text-to-speech and accessibility architectures are effective for African languages? | IDRC, FCDO | Easy Accessibility will build and validate accessibility infrastructure for languages with limited digital resources — an evidence gap that no existing research program addresses at platform scale. |
| How can results-based financing be designed so that outcome metrics are trusted, auditable, and resistant to gaming? | GPE, World Bank, FCDO | RBF4Ed's entire design — GEOS standards, GEOSor™ certification, anti-gaming constraints, reproducibility requirements — is a live implementation of the governance infrastructure that RBF requires. PREMIER's Easy Knowledge Assessment strengthens the assessment layer. |
| How should countries incorporate technology through curriculum reform and teacher support? | UNESCO (GEM Report 2023) | ECM's Curriculum IR + PROMISE teacher training + RESPECT Platform deployment generate longitudinal evidence on the interaction between curriculum digitization, teacher preparation, and learning outcomes — across multiple countries simultaneously. |
This table represents today's research frontier. As DPI-Ed scales — reaching more countries, more languages, more learners, and more years of longitudinal data — the frontier will expand. Questions that are not yet formulated will become tractable. The institute that can answer them will already exist.
PREMIER is the research engine of the Breakthrough System — and it is designed to endure. The founding portfolio — five Little Easies bundled with the Institute, plus two Big Easies (ECM and Easy FLN Localization) housed within the Institute and independently funded — addresses the most urgent shared capabilities for Africa's DPI-Ed. The institute's mandate extends beyond that founding portfolio, because the range and scope of hard problems that can be solved through applied R&D and then deployed at continental scale on Africa's DPI-Ed will grow with time. Every year of platform operation generates new data, new questions, and new research opportunities. The pipeline of tractable, high-impact research problems will expand as long as Africa's DPI-Ed serves learners.
PREMIER's embedded position within the RESPECT Ecosystem — with direct access to CRADLE's federated data and direct integration into the RESPECT Platform — compresses the research-to-deployment pathway from years to months, ensuring that every research dollar produces deployed infrastructure serving African learners. Development Partners funding PREMIER are investing in their own research agendas — through an infrastructure that makes the questions answerable for the first time.
A Development Partner that funds PREMIER's founding portfolio — the Institute plus the five Little Easies — is founding a research institution with a generational mandate, and holds Right of First Refusal on the two Big Easies that are the Institute's crown jewels. The Institute's contributions to African education, and to global education science, will compound for as long as Africa's DPI-Ed operates.