In Ecuador, 1 in 4 children under 5 suffers chronic malnutrition — often not from lack of food, but lack of guidance. Manka is a WhatsApp-native AI chatbot that reaches mothers with nutrition advice grounded in the foods they already have. Streaver helped Fundación ADHA take it from a hackathon MVP to a platform they own and run themselves.
Narcisa Ramírez lives in rural Tungurahua. She uses Manka on WhatsApp to decide what to cook for her baby with what's already in her kitchen. This is what she told ADHA.
Gracias a Manka me siento más segura de qué alimento puedo dar a mi familia, pero más a mi bebé — porque Manka me ayuda a solucionar alguna inquietud de algún producto que tengo en el campo y lo pueda consumir tranquilamente.
Thanks to Manka I feel more confident about what food I can give my family, especially my baby — because Manka helps me resolve any doubt about a product I have here in the countryside so I can use it with peace of mind.
Fundación ADHA is a social organization in Ambato, Ecuador, working directly with rural communities to address malnutrition. In Ecuador, 1 in 4 children under 5 suffers chronic malnutrition — one of the highest rates in Latin America. The cause, ADHA realised, is often not the absence of food.
Families in the communities ADHA serves typically have food available and even bring their children to health centers. They still feed them poorly, out of habit and a lack of culturally-grounded guidance — using what's in the kitchen well, knowing what to combine, knowing what's right for a particular age. Address water, hunger, or healthcare alone, and this cause of stunted child development goes untouched.
ADHA had a hackathon MVP of a nutrition chatbot called Manka. It needed to grow up — into something families could rely on, in low-connectivity rural conditions, in a way ADHA itself could own, operate, and grow after the build was done.
The hard part of Manka was never the AI. It was respecting four constraints at once — and refusing to treat any of them as a roadmap item to handle later.
The problem ADHA was solving isn't scarcity — it's guidance. Families have food. What they lack is grounded, culturally-specific advice about how to use it well for a particular child, at a particular age, in a particular season.
Knowledge gapMothers in the target communities don't have reliable broadband, don't carry the newest phones, and have varying levels of literacy. Any solution that asked them to install an app or follow a long onboarding flow was dead on arrival.
Real-world constraintsThe starting point was a hackathon MVP — a working idea, not a working platform. It needed to become something ADHA could operate on its own, monitor, update, and expand without depending on Streaver for every change.
MVP → owned platformGeneric nutrition advice misses. The guidance has to fit what's actually in mothers' kitchens, what's actually at the local market, and what's actually being cooked in those homes. Cultural specificity was a hard requirement, not a polish step.
Cultural specificityTech To The Rescue introduced ADHA to several candidates. Three reasons the match landed with Streaver — none of them about engineering capacity, all of them about context.
ADHA wasn't looking for a vendor — it was looking for a partner who shared the context. Streaver is a Latin American team that lives alongside these same realities. Spanish was only part of it; the team understood the mission from the inside, not from a brief.
Any solution had to work on WhatsApp, on real phones, in homes with patchy connectivity and varying literacy. Streaver had built communication systems on these constraints before and didn't have to be talked into respecting them.
The success criterion was ADHA's independence, not Streaver's retainer. The engagement was scoped end-to-end as a build-then-handoff: ADHA owns the platform, operates it, and grows it. The work isn't finished until ADHA doesn't need us anymore.
The shape of the work was a graduation. Take a scattered set of prototype pieces and turn them into one coherent thing — WhatsApp for mothers, a back-office for ADHA, an architecture that ADHA can keep running long after we're gone.
WhatsApp-native · ADHA-owned · scalable across rural communities
We didn't ask mothers to install anything. WhatsApp is already on the phones in these communities — already trusted, already understood. Building Manka as a WhatsApp-native chatbot removed the single biggest barrier to adoption and let the product meet families exactly where they already were.
We built the AI layer on AWS Bedrock — managed, in-region, and within ADHA's operational capacity. Self-hosted models would have given ADHA a science project to babysit. Bedrock gives them a service they can run with the team they already have.
Alongside the chatbot we built ADHA a web back-office: monitor conversations, manage guidance content, update advice as ADHA learns from the field. The back-office is what turns Manka from a piece of software ADHA was given into a platform ADHA owns and improves.
Twenty mothers in Tungurahua is a real, careful pilot — and it's still twenty mothers. We're being honest about that. The platform is built to scale; the proof that it scales hasn't happened yet, and we don't want to oversell what is, today, an early-stage social impact build.
Engagement metrics, message volume, and dietary-outcome signals from the pilot aren't measured yet. The qualitative response from mothers like Narcisa is strong; the quantitative story is still TBD and will be reported truthfully when it lands.
The single most powerful asset for this page is Narcisa's testimony — and right now it lives inside a longer ADHA / Tech To The Rescue film. Cutting a 20–40 second standalone clip of just her words would do more for this story than any new feature we ship next quarter.
Each outcome below is paired with the baseline it's measured against. Where the data is qualitative — and it mostly is, at this stage of the pilot — we've said so plainly.
The first cohort of mothers in Tungurahua is using Manka on WhatsApp today — getting real nutrition guidance, in Spanish, grounded in the foods they actually have.
Narcisa Ramírez's words — captured by ADHA — are the clearest signal we have that Manka does what it's meant to do: give a mother enough confidence to use what's already in her kitchen, well.
ADHA holds every credential, every repository, every dashboard. The team can update content, manage conversations, and grow Manka without us in the room.
The architecture, the WhatsApp-native pattern, and the back-office model become a template for other impact organisations doing similar work — in Ecuador and beyond.
Manka is Streaver's first Tech To The Rescue collaboration delivered end-to-end. It sets the model for how we engage with future impact matches.
The infrastructure is modular enough that SMS integration and adjacent health topics (maternal health, childhood development, water/sanitation) can be layered in without rebuilding.
From the very beginning, Streaver understood that Manka was more than a software project. They didn't just build what we asked for — they helped us think better and beyond. It truly felt like a partnership.
Four people did the work. Every one of them understood the brief — this is a build that ends with us walking away, and the platform still being healthy.
Owned the relationship with ADHA and the architectural calls. Sat with ADHA's founders during weekly syncs.
Built the Bedrock-backed AI layer and the retrieval pipeline that keeps Manka's advice grounded in ADHA's curated content.
Owned the WhatsApp integration and the conversational flows — making sure low-connectivity, real-phone constraints were respected.
Designed the ADHA-facing back-office so non-engineers can monitor conversations and update guidance content with confidence.
Manka is a Tech To The Rescue collaboration. Streaver was matched with ADHA pro bono / at cost via TTTR, and the engagement was scoped to graduate ADHA into full operational independence.
Build-for-impact pricing tied to handoff milestones, framed by the Tech To The Rescue collaboration. Streaver was matched with ADHA pro bono / at cost via TTTR — every milestone moved ADHA closer to operating Manka without us.
Weekly working sync with ADHA's founders, daily async on Slack. ADHA's voice was upstream of every product decision, not downstream of it.
Pricing moved on milestones tied to handoff: WhatsApp live, back-office shipped, pilot launched, full operational independence. Each milestone moved capability into ADHA's hands.
Full IP transfer and full operational independence on day 112. ADHA holds every credential, every repo, every dashboard. Streaver kept nothing that would make ADHA dependent on us.
Every choice in the stack was made through one lens: can ADHA operate this without a dedicated ML or DevOps hire? Managed services won wherever they could. We weren't building for ourselves.
Manka belongs to ADHA now. The roadmap below is ADHA's, not ours — the three threads they're pulling on next, with the platform we left them.
Manka is built so the same architecture can serve other rural communities — in Ecuador and across Latin America. The first replication conversations are underway, anchored to ADHA's network.
Not every mother has reliable WhatsApp. SMS is the next channel — same back-office, same content, broader reach into the lowest-connectivity homes.
The same rails can carry maternal health, childhood development, and water/sanitation guidance. Manka's success is the case for ADHA to expand the topics it serves on top of the platform it now owns.
Streaver partners with mission-driven organisations and impact funders to build technology that respects the communities it serves — and that the organisation can own, operate, and grow long after the build is done.
