Pinned Tweet
La semana pasada lanzamos Jelou Agent 2.0. Un agente que construye agentes de IA para conectar cualquier canal de comunicación (Whatsapp, email, voz, etc) y les permite ejecutar transacciones para cobrar con procesadores de pagos, validar identidad y transformar cualquier software que use tu empresa en un agente de IA.
Es una versión más potente para construir, y puedes usar los modelos que quieras en el cloud que quieras. Además, mejora tus agentes constantemente pues es capaz de analizar conversaciones, detectar problemas, mejorar flujos, ejecutar tareas durante horas y optimizar la operación.
Jelou Agent 2.0 también puede ayudarte a optimizar costos de tokens y mensajes de Whatsapp, generar reportes y construir desde donde ya trabajas: Claude, Cursor, Codex o nuestro CLI.
Yo estoy construyendo varios agentes desde el mismo Whatsapp con la integración de nuestro CLI, es decir un Agente en Whatsapp que construye otros agentes de Whatsapp.
La "inteligencia" es cada vez más accesible y poderosa; el reto es lograr que esa "inteligencia" hable con el software que ya usan los bancos, retailers y empresas de la región para acelerar su operación y distribución de servicios.
Empieza gratis en jelou.ai a construir agentes que van a generar ROI
Luis Loaiza retweeted
Desde octubre Meta cobra cada mensaje de servicio en WhatsApp 💸 Ahora el AI Agent en Jelou agrupa los mensajes del usuario y responde una sola vez: menos mensajes, menos costo y una conversación más natural 🦾👇 teapawt.org/luisloaiza/status/2122…
Users send "Hi", "I want to know your hours", "are you open Sunday?" as 3 separate messages.
AI Agent built on @JelouAI now waits, groups them, and replies once.
Configurable wait of 1–30 s, and messages sent mid-reply get queued, not dropped.
Off by default. AI Agent > Advanced > Group messages.
Users send "Hi", "I want to know your hours", "are you open Sunday?" as 3 separate messages.
AI Agent built on @JelouAI now waits, groups them, and replies once.
Configurable wait of 1–30 s, and messages sent mid-reply get queued, not dropped.
Off by default. AI Agent > Advanced > Group messages.
Ahora puedes ver tokens y costo de cada mensaje de tu AI Agent mientras pruebas.
Esto te permite comparar modelos y medir el impacto de un cambio en el prompt antes de publicar.
Al final de cada mensaje del AI Agent, bajo su última burbuja, aparece una línea con los tokens y el costo, por ejemplo 1.665 tokens · $0,0062. No hay que activar nada.
En el tester, CLI y en Whatsapp usando nuestro sandbox lo puedes ver en tiempo real mientras le hablas a tu agente.
WhatsApp: Nodo AI Agent → pestaña Avanzado → Mostrar costo al probar en WhatsApp
WhatsApp ahora cobra por mensaje. Entonces la IA también tiene que entender costos.
Jelou Agent detecta flujos caros, ráfagas innecesarias, prompts ineficientes y modelos sobredimensionados; luego propone y construye mejoras.
Menos mensajes. Menos tokens. Mejor outcome.
👋🏻 Say Jelou & Welcome to Cooperativa JEP 🇪🇨 ! JEP is now live with Emi, through our joint venture with Claro Empresas Ecuador
JEP is Ecuador’s largest credit union, the biggest cooperative in the country’s history. Founded in 1971 in Sayausí, Cuenca, by 29 young entrepreneurs, it now serves more than 1.5 million members, operates 64 agencies nationwide, runs 1,000+ ATMs, and ranks #1 among Ecuadorian cooperatives, #3 in Latin America, and among the top financial institutions in the country.
From a pre-approved offer to a configured and signed loan, all inside WhatsApp with orchestrated AI Agents.
AI connected to JEP’s core infrastructure, to run lending, cards, and collections in one conversation. On lending, members choose amount, term, payment date, and amortization type. Every change is recalculated live against the core banking system. Identity is verified with biometrics. The electronic signature happens in the same flow. No branch visit. No handoff.
Three financial processes in one intelligence layer, connected directly to the business. For banks and credit unions, this is the real shift: moving financial products from traditional digitalization to agent-driven execution.
Start building transactional AI for FREE: jelou.ai/en
Or book a call with our team to deploy transactional AI agents: lnkd.in/eZMU4a6K
Building transactional agents insanely fast.
In just a few weeks, we changed our entire AI deployment motion. What used to take weeks can now happen in hours. We’re even helping complete backend capabilities that customers haven’t been able to ship on time.
Today, we can deploy agents in as little as 24 hours, and these are not basic support bots. They can underwrite credit, trigger payments, connect to core systems, and execute real business transactions. Some customers are already seeing us build complex agents with them in real time.
This is also a lever for a new generation of AI deployment companies across the region. They shouldn’t need to charge Big Four-level implementation fees to deliver enterprise-grade AI. With Jelou, they can reuse infrastructure, local rails, integrations, QA, and transactional capabilities. One external deployment team, a Jelou builder partner, has already opened a $1M pipeline of opportunities using our harness. They can deploy faster, charge less, and compete for Fortune 500 customers with much better economics.
The bigger shift is that Jelou itself is becoming an AI agent: one that can understand what a customer needs, scope the solution and start building it on the spot.
Jelou Launch Week en vivo 🚀 ¿Puedes competir con Agencias de IA que conectan agentes a software corporativo?
Con una buena herramienta podemos hacer más: nuestros aliados pueden competir con nuevos players de AI deployment, a menor costo y con una ventaja clave, generar valor transaccional.
Estamos mostrando en vivo como migrar de ManyChat a IA transaccional en minutos. Lo mismo hacemos con infraestructura corporativa que aún no logra generar ROI por cada uso de tokens.
Nuestros partners y equipos de despliegue trabajan sobre un mismo "harness" para conectar sistemas, orquestar agentes y habilitar transacciones a las que clientes y usuarios pueden acceder sin fricción.
Queremos mostrar lo que se puede construir con Jelou AI: pruebas automáticas, QA, observabilidad y optimización de costos en tokens y mensajes, todo sobre una misma infraestructura.
Con este harness, cualquier empresa tecnológica puede desplegar IA con más velocidad, control y mejores unit economics, sin depender de una nueva capa costosa de servicios para cada implementación.
Pueden ver en vivo todo este proceso en este enlace: youtube.com/watch?v=-WIMG__P…
We’re saving hundreds of thousands of dollars on WhatsApp messaging costs to our customers!
In the video below, you can see how we do it using the Jelou AI CLI. You can work with Claude Code or directly from Jelou.ai and migrate expensive chatbot infrastructure in a matter of days.
We’ve already migrated infrastructure for large enterprises in <15 days, with a team specialized in agent deployment and focused on delivering results before October 1st (Whatsapp deadline to start charging per message).
If your current platform can’t migrate messages, flows, and integrations at the speed you need, we can help.
We build everything in a separate environment, validate the results, and switch you over only when everything is ready.
Connect with our team and use AI to lower service costs while improving customer outcomes: calendly.com/d/d3rf-sb5-bwy/…
Say Jelou & welcome to Banco Azteca. 🇲🇽 👋
Banco Azteca moved from a rule-based bot to an AI agent in production in just 4 weeks, securely connected to the bank’s infrastructure.
Together with our partner Dialogus, we transformed an existing foreign-exchange experience into a transactional AI agent on WhatsApp. Customers can check exchange rates, choose to buy or sell, select a currency and amount, confirm their quote, and find the nearest branch, all within the conversation.
By the time they arrive at the branch, the process is already underway.
An existing financial service, now running on a new layer of intelligence—built in weeks, not months.
For builders and partners, this is the opportunity: build and monetize transactional AI solutions for large enterprises using Jelou’s infrastructure.
Start building on Jelou apps.jelou.ai/signup
Want our team to help you deploy transactional AI with measurable ROI? jelou.ai/en/pricing#hs-meeti…
Your Voice AI can talk. But can it authenticate a customer, charge their card, or verify their identity to complete a wire transfer?
Most Voice AI today stops at the conversation. It can answer a call, understand intent, collect information, and trigger an API.
But a real transaction often needs more. Imagine a customer calling to make a payment. The Voice Agent handles the conversation, understands what they need, and stays with them throughout the process. When authentication or payment is required, it orchestrates with a WhatsApp Agent.
The customer receives the secure flow on WhatsApp, verifies their identity, and enters their card details through a PCI-compliant payment experience. Once completed, the Voice Agent already knows the result and continues the conversation. No starting over. No disconnected agents. Both channels share the same memory and business context.
This is the difference between adding Voice AI and building infrastructure where AI can actually transact.
Talk on Voice. Authenticate and transact on WhatsApp. One agent infrastructure behind both. Start with voice following the instructions here : docs.jelou.ai/en/guides/voic…
📣 Meta Business Agent is now available to try for free on Jelou.ai
You can now activate Meta’s native AI directly on your WhatsApp Cloud API number.
🧠 Configure it: define the prompt, add business knowledge, FAQs and skills, and connect workflows and tools.
🔒 Test with a controlled audience: create an allowlist of phone numbers so MBA only responds to selected users.
⚙️ How to activate it: Project Settings → Meta Business Agent → select your WhatsApp Cloud number → configure → activate.
💳 A new way to save on messaging costs as WhatsApp pricing changes in October.
Check more in docs.jelou.ai/en/changelog#m…
If you're building AI agents on WhatsApp, cost optimization is becoming part of the architecture.
Click to WhatsApp can help with acquisition. But what happens in customer service, where users initiate conversations and every extra message adds cost?
Builders need to optimize the conversation itself: fewer messages, smarter model routing, lower token usage, and visibility into the cost per outcome.
With Brain Studio, we're giving Builders the tools to measure and optimize exactly that.
In one healthcare use case, we reduced token and conversation costs by 70%.
Building the agent is only half the job. Building the right unit economics is the other half.
We want to welcome to Tempo Panamá as a new customer!
The experience of buying furniture and creating spaces for the home is becoming faster, more connected, and more scalable.
Tempo offers furniture and home solutions for bedrooms, living rooms, dining rooms, outdoor spaces, offices, entertainment, and décor. Learn more about Tempo at tempodesign.com.pa.
With Jelou AI, Tempo will manage customer inquiries about products, quotes, deliveries, installations, warranties, and repairs across WhatsApp, Instagram, and Facebook.
The goal is simple: reduce response times, streamline operations with transactional capabilities, and deliver a smoother customer experience across every digital channel.
If you want to start building experiences like these for your company or a customer, signup for free at apps.jelou.ai/signup?source=… and run AI agents built for transactions!
139
Meta Business Agent has been live and interoperating with Jelou AI for the past month. Here are a few things we’ve learned about this AI model, which changes the economics of WhatsApp conversations: instead of paying per conversational message, you pay based on token usage.
Meta Business Agent enables natural conversations, answers frequently asked questions, and can guide and recommend. What it doesn’t do, at least not yet, is execute: integrate with your CRM, process a payment according to your business rules and using local rails, verify identity using local government data, or provide an audit trail showing which system accessed which data.
That’s where interoperability comes in:
→ The customer sends a message on WhatsApp
→ MBA handles the conversation and detects an operational intent
→ It hands control over to Jelou
→ Jelou executes using the company’s rules, integrations, and traceability
→ MBA resumes the conversation with the result
For simple processes, MBA on its own may be enough. For processes involving data, payments, identity, or compliance, governed coexistence makes the difference.
Sharing your data
One advantage of interoperating with Jelou is that you don’t need to expose sensitive information to MBA—or any other AI model. Transactions don’t have to be executed by MBA, and you can build AI agents on our platform designed to protect user information while executing the underlying process.
Costs
When you use MBA for conversations with users on WhatsApp, Meta charges based on tokens rather than messages. Message-based charges apply when other models are used.
MBA’s token consumption can be significant: 5,000 tokens just to start a conversation, compared with 217 tokens when orchestrating models through Jelou.
Token usage per conversation or per outcome, directly affects margins per transaction, operation, and customer service interaction. Today, we help companies build high-quality conversational experiences on WhatsApp using multiple AI models, while controlling the budget per conversation and optimizing toward an outcome that generates ROI.
If a customer simply writes “Hi” and disappears, that interaction shouldn’t cost the same as another customer who completes an entire transaction.
Fixed pricing per conversation is disappearing, making way for a more transparent model focused on actual outcomes.
Customer support agents and automations are often fragmented, treating every customer the same.
Real personalization requires more than an LLM. It requires an agent that deeply understands each user's context, preferences, and history.
Every service agent should operate in concierge mode.
194
A bank as an app.
A bank as a service.
A bank as an AI agent.
Join us to build the next distribution layer for banking.
2
170
¡Ya se encuentra en producción el feature de Clonación de Workflows entre proyectos de una misma organización!
Esto nos facilita poder reutilizar los workflows que ya hayamos creado en otros proyectos sin requerir crearlos desde cero.
📷 ¿Cómo funciona?
Activa el modo de clonación en la barra lateral, en la sección Workflows, haz clic en el ícono de clonar workflows que está junto al +. La lista entra en modo de selección: cada workflow muestra una casilla y se ocultan los menús de acciones de cada fila.
Más detalles en nuestra documentación: docs.jelou.ai/guides/getting…
2
156















