Selling was always a conversation. Conversation makes conversion.

1Conversation was never replaced.

Each era added a new way to sell. None replaced the one before.

E-commerce solved several constraints of physical selling. It also moved buying from assisted to self-service, which goes against how many people like to buy. That's why e-commerce relies on driving traffic and why conversion is hard. Average store conversion sits around 1–3%.

Social commerce grew because it is assisted again, in a new form. People ask in a DM, comment, or join a live session, then buy with help.

SMEs sell a lot this way. Conversion is good, and people already spend their time on social media and find it easy to ask and buy in DMs. Bain observes higher purchase rates in conversational journeys.

Yet even with this potential, few brands default to selling through DMs. When they think of online selling, they build an e-commerce store first. Why?

Every assisted channel has one thing in common: conversation. And conversation makes conversion.

Chat converts. Chat does not scale.

2Chat does not scale.

Chat is not easy to scale. These constraints come from the channel itself:

• Replies are manual. Each "Is this available?" starts a one-to-one sale, so sales volume is tied to headcount.
• Orders are unstructured. Details are spread across messages and often confirmed by hand.
• Flow-based chatbots stop short of assisted selling. Menus, buttons and fixed paths automate replies, but the result is closer to an e-commerce click path inside chat than a real assisted conversation.

The conversion is real. The scale is not.

3Even a good chatbot is still a guess.

Building an effective chatbot takes more than building a chatbot. It usually takes two different skills — and even with both, a fundamental limit remains.

Skill 1 · Technical expertise

Someone has to know the platform: triggers, conditions, branches, integrations, and how to connect conversation steps into something that actually runs. They translate an idea into a working chatbot.

Skill 2 · Marketing expertise

Someone also has to know how customers buy: intent, consideration stages, objections, what information people need before they purchase, and how to guide different people toward a decision. This is often the harder skill. A strong marketer can design a sophisticated customer journey.

But even then, they are still designing journeys based on how customers are expected to behave.

Imagine an experienced marketer builds a chatbot for a laptop store. They research behaviour, map the funnel, and create careful paths for discovery, price questions, comparisons, purchase, order collection and after-sales. That is good marketing and good chatbot design.

It is still built around common scenarios. Individual customers bring different questions, priorities and circumstances. Conversations change direction. You can add branches, conditions and personalisation rules — and modern flow builders can do a lot of that — but you are still anticipating and programming possibilities beforehand. The more you try to cover, the more you have to maintain.

You can cover most journeys. You cannot personally design every one.

That is the gap. Ordinary business owners should not need to be both a chatbot developer and a marketing strategist just to automate customer conversations.

There is a simpler way. With Chatouts, you do not build conversation flows or map every possible path. You add your products and business information, choose your AI agent, and decide what it can do. The agent adapts to each conversation. Chatouts handles the complexity behind the scenes.

You configure your business, not your customer journey.

4AI cannot do this alone.

For the first time, software can hold a real assisted conversation: answer, recommend and take the order. But AI alone isn't enough. It needs a platform around it, with structure, accurate seller data and guardrails.

To work as a sales agent, that platform has to pass three tests:

• ChatGPT-level, on your data. It converses fluently, but only about your products, services and prices.
• Never a wrong price. One incorrect quote can cost a seller a customer's trust.
• Cost below the alternatives. The cost per conversation should stay below the other ways of getting the same sale. See the comparison below.

The tests pull against each other. Sending the whole shop with every message raises cost, and a lighter model can lower accuracy. Balancing all three is the design challenge.

The cost must be lower.

A business can get the same sale in three ways. Each has a different cost structure.

If it costs more than staff, or more than ads, it does not exist.

5The store belongs in the conversation.

Most brands already spend large ad budgets on social media. That's where customers spend their time, the crowded place. Yet DMs are usually treated as support and lead capture.

Opening a store where the crowd already is has always made sense. Now imagine infrastructure that sits inside your DMs and passes all three tests:

What if you turned your social DMs into your store, and sold with this infrastructure?

If conversion doubles on the traffic you already have, that is a second ad budget you do not spend.

Why that would matter. The usual way to double results is to double ad spend, buying twice the traffic. If conversion in DMs doubled with the same traffic and the same effort, it would deliver what a whole second ad budget delivers, without spending it. Tripling would equal two extra budgets.

That is the test. Conversation makes conversion.

6That is why Chatouts exists.

Chatouts exists so every business can hold the conversations that make conversion, at scale — without becoming chatbot engineers or journey designers first.

At the centre is an AI sales agent. It understands what customers need, recommends products, answers questions, and takes orders. Merchants spend less time managing conversations and more time growing the business.

Simple above the waterline, serious below. The complexity stays inside the platform. For sellers it's as simple as a Facebook Page, and for their customers every answer can be trusted.

• N-Engine decides what to answer, when, and under which conditions. It calls the AI model only when needed, which keeps pricing accessible.
• Memory and Context keep what the customer has already said, so the agent can adapt as the conversation evolves — without a predefined flow.
• Guardrails check every commerce fact before it reaches the buyer. Price, stock and order facts stay true.

We borrow frontier AI. We own the guardrails, Memory and Context, and N-Engine.

The agent adapts the conversation. The platform still controls execution for order-taking and other business-critical steps — with grounded facts, clear permissions and deterministic workflows where reliability matters.

We start before payment and delivery. Across Southeast Asia, cash on delivery, bank transfers and couriers already serve sellers well. The constraint sits earlier, in the conversation and the order, so that's where Chatouts begins: turning conversations into structured orders automatically.

What we believe

• Businesses should not need to become chatbot engineers to sell through conversations.
• Merchants configure the business — products, information, policies and what the agent can do — not conversation paths.
• Do not make businesses predict how customers will behave. Build systems that understand how customers actually behave.
• Intelligence and execution belong to the platform. Sellers stay focused on the business.

Your business. Your agent. Our intelligence.

Our principles

• Accuracy before automation. Every answer is grounded in the seller's own data. Price, stock and order facts stay true.
• As simple as a Facebook Page. No code and no flows. Add your products, business information and policies — the agent learns how your business operates.
• AI does the work, people handle exceptions. The seller stays in control.
• Start where the constraint is. Conversation and order structure first; payments and delivery partnerships follow.
• Accessible cost. An AI-assisted conversation only makes sense if it costs less than the alternatives.
• Bounded autonomy. Dynamic does not mean unconstrained. Adaptation sits inside permissions, validation and reliable execution.

7Where we're going.

Markets. First Myanmar and Cambodia (8.55M and 8.60M Messenger ad reach), then the Philippines (61.8M), then WhatsApp markets.³

—A note from our founder

For eight years I sold through DMs, answering "how much?", recommending and closing orders one chat at a time. Along the way, I studied marketing to CIM Level 7. Both taught me that many people value help when they decide.

E-commerce is mature and serves buyers well, and many people also buy through Messenger, Instagram and live streams, because they like to talk to someone. Chat selling works. Its constraint has been scale.

AI is the window to change that, if it has the right platform around it. With hands-on selling experience and AI together, we can help this channel work at scale — without asking every seller to become both a chatbot builder and a journey strategist.

Building a company takes more than one person's skills, and I'm learning every day. So we're building Chatouts with the sellers who use it. If you sell in chat, I'd love your help shaping it.

Ye Min Thu · Founder, Chatouts

—References

1. Shopify, Ecommerce Conversion Rate Benchmarks (2026)
2. Bain & Company, Win with Conversations (2024), Figure 7
3. Meta ad reach, DataReportal Digital 2025 (Jan 2025)