Blog

As discussed in a previous post, the basic version of an LLM-based advisory bot for MSMEs is like some all-knowing guru that sits in your WhatsApp contact list and waits for questions. This could be based on the LLM's training data, or on top of management best practice guides: a 'chat with your training manuals' tool.This places all the burden on the entrepreneurs to know the space of potential questions and use cases. It also places the burden on them to come back to the tool despite all the demands on their time and attention. And it does not help them at all with the most important step of translating information into an action plan, and following up. If a real life business coach did that, they wouldn't be very helpful, and we can see that an LLM-based tool working like that is also not very helpful and is quickly forgotten. Since 2023, WhatsApp has been running <a data-mce-href="https://developers.facebook.com/docs/whatsapp/cloud-api/guides/experiments/" href="https://developers.facebook.com/docs/whatsapp/cloud-api/guides/experiments/" target="_blank" rel="noopener">an experiement</a> and a random selection of 1% of users can't receive any business messages unless they write first. We have a few of these accounts in our user pool, and find that they ask a few questions when they first try the tool, and then go back to life and all the other demands on their time and rarely re-engage.&nbsp;To be useful, there needs to be a two-way communication, rather than just a guru sitting on a mountain. In the other implementation of this type of tool <a data-mce-href="https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4671369" href="https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4671369" target="_blank" rel="noopener">in Kenya by the Otis et al team of academic researchers</a>, their bot sent out reminders to use the tool (note though that those participants were pre-screened for the study and offered a reward for their completion, so its pretty different from a real-world set-up). There are many other promising ways that these tools can proactively engage with MSME entrepreneurs:- Send out 'tips of the week' based on training materials- Share the most frequently asked questions by other users from the previous month- Scope out well-defined use cases where the tool can offer more direct and applied advice and even help complete tasks (eg upload a photo of your display for step-by-step imrpovements, calculate and evaluate margins on a particular product to plan a sale, get a new logo, etc)- Exercises that the entrepreneurial training research has shown impactful for MSMEs (eg a goal setting module, a kaizen-type module, personal initiative-focused module)- Personalized follow-up to build momentum (eg last week we talked about X and you said you wanted to try Y. Did you try it? How did that go?)A core focus of our work is testing these and other ideas out, and finding out from our growing set of Bodega owners which ones were must useful and impactful for their businesses, and how else they'd like the tool to help them. Becuase they want and deserve something more useful than 'chat with the training manual'.

Beyond 'chat with the training manual'

For many applications, 'user engagement' is their north star. It makes sense: users have to actually use your product, and the more they use it, the more useful it is, right? This is clearly important if you are selling ads, but what if you are offering advice. Is more advice better?I spoke with a MAIA user recently. After what looked like some useful discussions with the tool around how to improve sales (store appearance, customer access, etc), she did not re-engage for nearly a month. I took that absense as an indication that she didn't find the tool useful, and wanted to learn more. But she said quite the opposite: she loved the tool. She was in the process of implementing a handfull of changes based on her conversation, and was already seeing very positive results. Some of those changes involved physical changes (moving gates) that take some time. I asked why she wasn't using MAIA regularly to find more improvements, and she said that she will come back to it when she needs new ideas, but right now she's still implementing the first action plan. What's the point of more ideas if she doesn't have the time to actually enact them?I hate golf, but its a useful analogy. If you take a golf lesson, a bad pro will give you 20 generic tips to improve your swing, you'll remember one or two of them, and maybe you'll get a bit better. A good pro, however, will give you just the one tip that will fix your most critical problem at that point in your evoluation as a player, for maximum impact. The same holds for business advice. Better consultants don't give more recommendations, they give the smallest set of key improvements that have maximum impact.So we will look at and report on engagement. But beyond some point, more usage isn't necessarly better, it could actually be worse. So business performance and improvement will be our north star, not engagement, and we'll figure out what engagement patterns are associated with maximum business impact.[PS- the day after posting this, I attended a presentation by Nick Otis, the lead author of that interesting Kenya paper mentioned earlier. He mentioned that in his recent research, they've found some negative relationships between intensity of AI mentor usage and business performance. His explanation for this was that chatting with the tool could at some point substitute for talking with real people, which has some benefits that AI doesn't eg introductions and networking. This is analogous to what researchers find about why screen time is a problem for young kids]

Engagement, More Advice & Better Advice

In a future blog post I will analyze these questions as well as provide some conversation logs, but first we want to make the raw data available. As per our commitment to users, all personally identifiable information is redacted (****), and the rest is made available to the research community for R&amp;D purposes without the ability to identify or contact any user. Questions are provided as originally written, unless transcribed from a voice memo. This is a complete list of questions over the past few months from identified users (as opposed to system testers from partner organizations, the topic of a future post), only those from the small-scale retail sector. If you have any comments or insights we’d love to hear about them at mype.asesor.ia.sac@gmail.com[these are user-initiated questions. So it excludes all user questions in response to system-generated messages and tips. It excludes all follow-up questions on the same or similar topic within the conversation that the initial question started] USER QUESTIONSQué está pasando con el movimiento económico? ya que es mi punto es el negocio has bajado muchoRobo ormiga&nbsp;Me gustaría abordar el tema de la fachada o frontis de mi local me gusta me gustaría que sea más profesional y que llame más la atención para las ventasBuenos días mi principal problema es el maltrato al cliente yo estoy en todas las áreas de la bodega recepción de pedidos atención al cliente y atención de proveedores **** la bodega atiende los 7 días de la semana 15 horas diarias lo cual me desarrollado estrés **** lo cual me lleva a tratar mal al cliente en muchas oportunidades durante el día graciasComo calificar a un asistente para la bodegaTengo más gastos que ingresos. hace poco me hackearon mis cuentas bancarias haciendo que pierda cantidades fuerte de dinero por lo cual estado trabajando con tarjetas de crédito para comprar mercadería y las ventas estado muy bajo por mi zona afectando mi economía e incluso ya agote los préstamos personales de amistades y familiaLo que quiero saber es como administrar mis ventas. La verdad que yo vendo y apunto en un cuaderno y con el dinero compro lo que necesito de mercadería, pero a veces me falta.&nbsp;Cómo podría vender más?En el punto de Mi Local Ps Tengo Mucha competenciaCómo puedo acomodar mejor mi Bodega, mi local es amplio pero al FondoEs Bueno vender pan en la BodegaCómo atraer mas clientes a mi BodegaLa **** crédito me está vendiendo muy caroClaro, qué tal sabes el problema qie tengo es que tengo muchas deudas y quiero aumentar mis ventas por catalgo y no se como enpesarComo puedo mejorar mis ventas&nbsp;Quiero controlar mis inventarios y mi control de crédito a mis clientes y proveedoresQuiero saber como puedo hacer marketing para que se vea más vistoso mi bodegaQuiero incrementar mis ventas pero no sé como hacer tiktok ni estados de WhatsApp para promocionarmeQuiero vender más , mis ventas han bajadodebo vender todo a precio sugerido?Vender masTengo una bodega y quiero hacerla crecer y vender más, que herramientas o acciones me recomendarías hacer? Quiero convertirla en un minimarket o algo mas exitoso.Cómo puedo ordenar las mercaderías en mi negocioQue se le puede dar por el día de la madre??Está es mi bodega, que debería modificar para que se vea más ordenada. Que color de paredes me recomiendasComo puedo ganar tengo una bodega y las ganancias no los veoY que me recomienda para para llevar mis ganancias y gastos k aplicativoA veces las ventas son bajas en días normalesComo mejorar en las ventasComo puedo llevar la contabilidad de mi negocio, saber cuanta ganancia tengo, es un negocio pequeñoQuisiera mejorar mis finanzas y aprender más de administraci⁹nCómo puedo manejar mi stock y reducir la mermaComo mejorar la atención al clienteCómo puedo mejorar mis ventas usando las redes ?Quisiera vender mas y gastar menosComo organizar la bodegaCómo mejorar mi inventarioSobre ofertasComo exhibir los productosQuisiera saber como hacer crecer mi negocioHola tengo una bodega y no se coml organizarme para hacer los inventarios y el flujo de caja diario.Tienes alguna distribucion para una bodega de &nbsp;4 x 6 metrosComo lidiar con clientes malcriados ?Hola, en mi tienda trabajamos solo dos personas y el contador. Como puedo armar mi organigrama y cuales serian nuestras obligaciones?Tengo una pequeña tienda, cómo puedo generar más ventas?Quiero abrir un nuevo negocioQuiero potenciar mis ingresos mensuales mediante ventas a larga distanciaQuiero saber aplicar un layout en mi tienda abarrotes de 49 metros cuadradosUbicacion de &nbsp;mis &nbsp;productosEl nombre de &nbsp;mi &nbsp;negocio es &nbsp;**** QUE OPINION &nbsp;TIENE. ESTA BIEN ESE &nbsp;NOMBRE &nbsp; O DEBO CAMBIARLO?Kiero &nbsp;vender &nbsp;otros &nbsp;productos &nbsp;.en la diversidad de &nbsp;productos. Kiero vender &nbsp;utiles escolaresMe &nbsp;puedes &nbsp;ayudar &nbsp;con &nbsp;el &nbsp;logo de mi bodegaMe &nbsp;puedes &nbsp;explicar &nbsp;que es el impusto selectivo al consumo &nbsp;. Y &nbsp;si esto aplica a mi negocio. Osea &nbsp;¿ si vendo cerveza yo &nbsp;tengo &nbsp;que &nbsp;pagar este &nbsp;impusto ?Mis &nbsp;percepciones &nbsp;no me devuelven aun &nbsp;solicite &nbsp;mis &nbsp;percepciones me &nbsp; dieron un &nbsp;monto a &nbsp;cobrar &nbsp;y me dijero que en enero me darian y hasta ahora &nbsp;nada &nbsp;no me &nbsp;depositan. Que &nbsp;debo hacerDe &nbsp;manera, &nbsp;como &nbsp;cultura &nbsp;general &nbsp;. ¿ que es una &nbsp;empresa &nbsp;hibrida ?Hola, sobre mejorar el ambiente en mi bodega. Acomodo de mercadería. Finanzas y estrategias para promociones. Gracias por tu apoyo.Necesito subir mis ventas ya que está bajando demasiadoEl rendimiento es un poco bajo porque existe bastante competencia cerca, el problema seria la competenciaComo mejoraria en mis gastosComo hacer Promociones??Como seleccionar a un personal de apoyo colaborador?Que es Merchdaising

What do users ask?

Junior is a bodega owner in Peru who shares tools and tips with his fellow bodegueros on TikTok. After trying out Maia, he was kind enough to make and share the following video with his 1757 followers:<a href="https://vm.tiktok.com/ZMMVDC7qy/" data-mce-href="https://vm.tiktok.com/ZMMVDC7qy/"> https://vm.tiktok.com/ZMMVDC7qy/</a>His followers were told to comment "IA" on the video to receive a link to access Maia via direct message from Junior. As of April 29th, we could see there were 284 comments requesting access on the video. Not all of those had accounts that could receive DMs from comments, and we're not sure who Junior replied to, but in the 48 hours after his post, 28 new users accessed Maia.&nbsp;1757 followers -&gt; 284 expressed interest -&gt; 28 were connected to maiaOf those 28, 3 did not engage with the tool, and of the remaining 25, there were an average of 6 messages exchanged. One of those 25 is enrolled in the WhatsApp experiment group meaning they can't receive messages. The remaining 24 were sent a tip of the week, and then the following week, were asked with a quick-response poll question if they wanted to continue to receive the tip of the week or unsubscribe. Of those 24, 15 replied that they want to continue their enrollment, 0 replied no, and 9 did not reply (response rate of 63%).

Statistics from an Influencer's Post (Updated)

The average Bodega owner is in their 50s, so thinking about entrepreneurial training in this sector is not about pedagogy, it is about andragogy: self-directed learning for adults. And a big difference in andragogy vs pedagogy is the importance of relevant, contextualized learning. Adult learners don't have the same interest in (or patience for) abstract, theoretical explorations. Learning for adults is more effecive when it is problem-centered, immediately relevant, and experiential.&nbsp;This is clearly an advantage of continuous AI-driven entrepreneurial training versus up-front classroom-based training. Tools like MYPE AsesorIA can serve up knowledge in an interactive way that is immediately relevant to whatever problem the entrepreneur is facing in that moment, incorporating context. This is much more in tune with how adults learn.I also happen to be an adult (even moreso than 25 years ago) and the same applies to me: understanding comes through quickly moving from theoretical ideas to practical examples immediately relevant to what I am working on and most interested in. Which brings me to <a data-mce-href="https://en.wikipedia.org/wiki/Intelligent_agent" href="https://en.wikipedia.org/wiki/Intelligent_agent" target="_blank" rel="noopener">AI Agents</a>. Like most people working with AI, I've heard talk and excitement about AI Agents for over a year now, but didn't pay so much attention. But I've been thinking about them more now, because of conversations like this:-------------------Paraphrased chat between Celinda (a small-scale coffee grower in Peru) and Maia (Mype AsesorIA)Celinda: my coffee is real good, and I get a lot of good feedback about its odor and flavor. Right now we sell whole-bean to aggregators but want to sell packaged, ground coffee, because I could earn a lot more.Maia: that's a great idea. With ground coffee, maintaining that odor and flavor has to be maintained with the right kind of hermeticlaly sealed packaging. How are you packaging your ground coffee right now?(Maia and Celinda have a back-and-forth about current packaging, the impacts of light and humidity on freshness, etc. and figure out the detail of exactly what packaging Celinda wants)Maia: to find packaging that meets these criteria at a good cost, you can talk with some of your fellow cafetaleros, co-ops, and suppliers, and search online. Here are the steps to search online (gives steps for contacting suppliers, giving requirements, collecting and negoating prices).Celinda: thanks. But what I would like is for you to do that for me. Can you?-------------------The step-by-step detailed advice offered here was great, and is surely a boost to the productivity of MSME owners like Celinda. But she wants more than just detailed, step-by-step advice. She wants the system to actually get out there into her real world do some work for her. Most AI agent discussions are about things like autonomous programming tools and virtual travel agents, working away in the digital world. But what would AI agents look like for the hundreds of millions of analog MSMEs in developing countries? What is a solution for Celinda where the AI doesn't just give detailed advice, but could operate on her behalf, collect specs from a variety of suppliers, and negotiate prices for her (which in Peru would require not only web serches and emails, but phone calls)?

Andragogy & AI Agents for Analog MSMEs

A super interesting study <a data-mce-href="https://twitter.com/emollick/status/1770618237782307075" href="https://twitter.com/emollick/status/1770618237782307075" target="_blank" rel="noopener">as summarized by Ethan Mollick</a>: "Bloomberg trained a GPT-3.5 class AI on their own financial data last year… …only to find that GPT-4 8k, without specialized finance training, beat it on almost all finance tasks. Hard to beat the frontier models."Many organizations are applying AI in International Development, and have sensible arguments for fine-tuning their own models. A particular country's language may be poorly represented in LLM training data, the application may be very sepcialized (eg agricultural extension), there are government rules around where data goes, and a fine-tuned open source model may look much more attractive when making financial projections out to millions and millions of users.&nbsp;There are many valid reasons, but we should also be on guard for the risk of some motivated reasoning here: non-profits still have incentives to 'build a moat' to survive and thrive. An NGO working to be a leader in AI and _ will, just like many AI start-ups, not want to look like a thin GPT "wrapper", and one way to do that is to fine-tune your own model instead of using off-the-shelf models.&nbsp;We are taking the opposite approach, and trying to be the thinnest wrapper possible that achieves significant impact for MSMEs. Much like unconditional cash transfers give an impact benchmark that more highly-engineered development interventions must beat, our results give an impact benchmark that proprietary models should beat, and which any institution can quickly implement on their own and achieve. Its important to avoid spending donor money to train another BloombergGPT.----------An aside on the other issues raised above:COST: Yes its true that GPT4 is expensive now, GPT4V has a low rate limit, etc. But look at the prices of GPT 3.5 a year ago compared to now, and project prices forward the 12-18mos when a solution is fully tested out and optimized, before which it isn't ready for massive scale-up.&nbsp;LANGUAGE: To figure out the right intervention model and UX we're starting with what is a very easy language (Spanish) that still gives access to millions and millions of low-income MSMEs needing help.&nbsp;SPECIALIZATION: The services sector absorbs a massive amount of labor in low-income countries, with a generic production process that is relatively easy for LLMs to address, and persistent productivity gaps, so its the obvious place to start. But we will test how specalized a production process has to get before the simple approach of a frontier model supplemented with basic RAG doesn't outperform a less capable fine-tuned model, potentially in the coffee sector.

Blog

Beyond 'chat with the training manual'

Engagement, More Advice & Better Advice

What do users ask?

Statistics from an Influencer's Post (Updated)

Andragogy & AI Agents for Analog MSMEs

BloombergGPT, Fine Tuning & Moat building