Artificial Intelligence Projects

Our goal and strategy

In a world of information abundance, Project ECHO aims to use AI to transform knowledge into wisdom through contextualized support. The ECHO Model has always been about people learning from people; AI is a tool that helps that learning reach further, not a substitute for it.

We are actively researching and developing innovative solutions leveraging the power of A.I. for both internal and external use, ensuring a responsible, safe, and ethical implementation, grounded in our Resonsible AI Governance Principles.

Our implementation of AI is broadly in 3 categories:

1. Integration of AI technology (e.g., Large Language Models, Agent Harnesses, Vector Databases) into the ECHO model and iECHO to enhance the user experience for ECHO partners and participants
2. Use of AI tools by ECHO staff to augment and automate workflows and improve efficiency (e.g., Coding, Testing, Automating research synthesis, data reporting etc.)
3. AI as a "listening engine" across the ECHO community that synthesizes learnings and trends from across the global network

Current implemented AI projects

iECHO Program Banner image generation Live in production

Technologies used: Google Gemini, Nano Banana
When a new program is created on iECHO, iECHO uses A.I. to suggest a banner image that is generated keeping in mind the program's focus area and topics, to improve the look and feel, and engagement of the program pages

PHI & PII detection infrastructure Internal beta testing

Technologies used: Google Gemini
When content is uploaded on to iECHO, the internal PHI and PII detection algorithm will check if there is any accidental disclosure of patient details, and flag it to the user, improving iECHO's data privacy and security compliance.

iECHO Program name suggestion Live in production

Technologies used: Google Gemini

When a new program is created on iECHO, iECHO uses A.I. to suggest suitable program names that are aligned with Project ECHO's program naming convention, thereby improving the quality of program names, and ensuring participants can easily find and register for programs.

ECHO Research Synthesis Agent Internal testing

Technologies used: Google Gemini, Anthropic Claude

Multi-step reasoning and synthesis agent, trained on 900+ peer reviewed publications about the ECHO model, designed to be a useful internal tool for key stakeholders to synthesize research and apply it for fundraising and other strategic communication to demonstrate impact in different regions, and against various aspects (e.g., cost effectiveness, efficacy, patient outcomes, scalability, sustainability, etc.)

Available at: https://about2.iecho.org/research-synthesis

iECHO Support Agent Internal testing

Technologies used: Google Gemini, Anthropic Claude

User support agent available in 25 languages that is trained on iECHO official documentation and best practices, designed to assist ECHO partners and participants in their queries and troubleshooting related to iECHO

Active Development

We are conducting active Research & Development in the following areas:

• Session Assistant: Summarizing learnings from ECHO sessions into clear takeaways - that can be shared and queried by participants in case they missed the session, or want to refer back later on
• Learning Advisor: to connect participants with suitable ECHO programs that align with their goals / focus area / profession and guide participants in their learning journey
• Analyst Agent: Analyzing data, enabling M&E and generating insights for hub teams - as well as making it easier to query for reporting metrics, insights etc - instead of having to do complex analysis in Excel. A secure and data-privacy-compliant AI agent is being developed that will be able to summarize de-identified data and generate trends and insights for partners and program teams.
• Simulation  / AI generated cases: that can supplement (not replace) participant-driven cases (we want to be careful not to use AI to reduce peer to peer interaction)
• Knowledge Engine: Analyzing insights and learnings from across the ECHO network to understand real time health trends / insights

Internal tool usage at ECHO

We have begun active pilots of utilizing tools like Google Gemini, Claude Code, and Copilot to enhance operational workflows across teams. Below are a few use cases that we are currently piloting

Foundational infrastructure and key components

Our goal is to build a re-usable AI stack that seamlessly integrates with the iECHO digital infrastructure, across the 3 layers of the platform, along with cross-cutting services handling all aspects of governance, security, observability and oversight.