Plan for when it breaks

"Ok, fine. But we won't buy that." That's what farmers kept saying when presented with the tractors the best experts in the world said they needed. Jehiel Oliver from Hello Tractor talks about how learn to listen to the market first, and the experts later. That's especially true when you're dealing with farmers, who know that no matter how ambitious a startup founder you are, you still can't break the laws of physics. The only signal that matters is whether the actual user will adopt the solution in their environment. A product can be technically elegant and still be irrelevant. If you optimize for what prestigious outsiders recommend, you can burn years building something the market quietly refuses to buy. He also gives real lessons on what it takes to lead: start from humility. Don't force yourself to move faster than a market will scale because you think you can solve all the world's problems. Make space for your teams to admit failure by admitting failure yourself. Spend at least two days a week making sure your teams are on track, and find ways to incentivize success and creative solutions.

The smartest move in AI today is to evaluate what works before you scale on excitement alone. What happens when Low and Middle Income Countries start pulling investments away from traditional health, judicial, or extension systems in order to build AI solutions instead? It can have huge upsides, but if you're only testing the AI and not the impact for people, it could lead to "irreversible losses." Listen to Attaullah Abbasi from J-PAL talk about their new AI Evidence Playbook.  AI is moving faster than the evidence can keep up, and that means people are often making decisions about if an AI "works" with engagement metrics like user adoption and message volume. Those can be dangerously misleading. The success of many AI tools drops dramatically when you release them into the wild, because if people don't have the bandwidth to use them well, or the system can't act on what the AI found, your investment isn't delivering real change. Atta talks about the difference between the speed of evidence, the speed of technology, and the speed of change. Your AI tool might evolve hundreds of times weeks, but it's still going to take a whole school year to see if kids passed the end of year reading exam. The evidence playbook gives you practical tools to understand what is working and what's not, and what methods will give you good enough information to move to the next phase of investment.

Cocoa income comes just twice a year, but families need stability every day. Carol Miloky (CARE) and Fatou Ndiaye (Mars) unpack what they learned from a decade of working with cocoa‑growing families. Savings groups were helping women build businesses, but without stronger household planning and clearer financial decision‑making, those gains didn’t always translate into lasting resilience. Over time, CARE and Mars saw a simple truth: stability at home shows up as stability in the supply chain. They explore how focusing on whole‑household financial visioning, helping families plan across the year, smooth income gaps, and make shared decisions, creates more predictable livelihoods and stronger supplier relationships. The result? Stronger enterprises, safer homes, and, as Fatou puts it, farmers who stay committed for the long term. Read more about this decade-long partnership here Women for Change - Empowering Women Cocoa Farmers - CARE

Amanda Larson and Aimee Mateo talk about building health technology in the Philippines. HEAL Hub is an app that helps local health workers stay up to date on training and the materials they need. But the health workers aren’t who you think they are: in many areas, health workers are on average 60 years old, and handing them a smart phone is not going to immediately translate into results, no matter how good the app is. Amy says the biggest change she would make is to start earlier designing WITH health workers, to understand what they need, what the Ministry of Health requires, and what kinds of examples will make a difference for them. Their biggest goal now is to figure out how the tool can integrate into the health system so it can last for the long haul.

Ask yourself, 'Did the system become stronger because we were there?’” That’s the biggest piece of advice Nelima Otipa and Nithya Ramanathan from Nexleaf Analytics have for you about trying to scale up technology. It’s not about building the perfect tech. It’s about seeing people as actors in their own systems. You need to adapt with governments, not sell to them. Inspired by their article on Scale that Lasts in Stanford Social Innovation Review (SSIR), we talk about why governments often prefer paper systems, how reliability is the key to trust, and what it takes to build a culture where you lead by understanding failure instead of ignoring it.

What happens when your brand new tool makes things worse for low performing entrepreneurs? Or restricts your most successful teachers so they can't unlock the full power of their skills? One of the most important questions you can ask about an AI tool is NOT, "is it working?" You really need to ask, "who is it working for?" and "How did the change happen?" Crystal Huang from IDinsight and Matthew Smith from IDRC talk about ways to evaluate AI tools, and how to think about if they really are creating the change you hoped for. Just like the internet didn't end poverty, neither will AI. What succeeds or fails will depend on how well we implement AI, and whether we can truly center how technology connects to humans and real life contexts. What do we do to take something that works in a controlled setting and help it survive the messiness of real-world implementation?

Robots that get stuck in the mud, a "successful" product a farmer will never use more than once, and a financial model clients can never pay for. Listen to Ankit Chandra and Ishani Lal talk about their article, Why AgTech Startups Fail, what inspired them, and what they learned. The core lesson is that we need to change what counts as success in AgTech. Success is not "does it work in the lab?" Success is, "Will this save the farmer time or money? Will it increase profit and lower risk?" Maybe most importantly, "will any farmer ever choose to use this more than once?"

Lindsey Moore was working in AI before most of us knew what it was, and she can tell you the most common mistakes to avoid. Ignoring context, building ever more precise models that provide terrible answers, and assuming that AI will replace smart strategy and human decision-making are three on the top of her list. If you're looking to do more with AI, she recommends you invest in learning good research methods, double-down on your data architecture, find ways to counteract bias, and stay skeptical. Developmetrics' Large Language Model with was trained on 60 years of USAID documents, and taps into a wealth of expertise that doesn't exist anywhere else. It can tell you what has worked, and what hasn't, over decades of work in dozens of countries. Here's what it tells us: we often repeat the same failures over and over again. Why? Because failures are as much about organizations as they are about tactics. The newest widget won't solve an organizational culture that drives people away from spending time understanding the local context. 

Eric Kaduru and Julia Arnold talk about why simply distributing phones doesn't help people--especially women--access the internet. After seeing free phones get broken, stolen, or cause men to punish women for owning phones, they needed a new plan. Instead, they talked about learning from HIV prevention campaigns in the 90s, demystifying something complex, and making learning accessible. Social norms are at least as important as technology. Soap operas, hip hop concerts, and talking to men are critical tools in opening up access to tech for women in Uganda, and worked better than free phones ever have.

CARE has a more than 30 year history with savings groups--starting from the lowest tech version you can possibly imagine: 25 women with a box and a notebook in Niger. Building on that, in 2013, we launched the process of building Chomoka--an app that would help women in savings groups manage their record keeping and connect to digital finance. Christian Pennotti talks about that journey, and why we finally passed the work over to Ensibuuko. Building a successful, scalable enterprise around digital tools was ultimately better put in the hands of a tech company that focuses on banking tech for people who can't access existing tools. What did we learn, and what are we doing now? Tune in to find out.