The phone is still the front door for a surprising number of businesses.
People call because they want an answer now. They want to know whether a service is available, explain a problem, change an appointment or talk to somebody before deciding what to do next. The interaction is immediate and familiar. It is also difficult for a small team to handle consistently.
That is the starting point for Mon Réceptionniste.
The problem is not “missed calls”
“Never miss a call” sounds like an obvious promise, but it is too shallow to guide a product.
The real problem is a broken connection between a person asking for help and a professional already doing their job. Picking up the phone can interrupt the work in front of you. Not picking up can mean losing context, trust or an opportunity. Traditional answering services can cover part of the gap, but they are not always available, specific or connected to the rest of the business.
The product has to create a useful outcome on both sides. The caller should be understood and directed. The professional should receive structured, actionable information rather than another inbox to process.
Voice AI changed the design space
Until recently, an automated phone experience usually meant a rigid menu: press one, repeat yourself, wait, start again.
Modern voice models make a more natural conversation possible. But a natural voice is not a product. The difficult questions begin after the demo works:
- What should the agent know?
- When should it ask a follow-up question?
- Which actions can it take safely?
- How does it hand the conversation back to a person?
- What happens when it is uncertain?
- How does the business understand what happened afterwards?
Those questions sit across product design, conversation design, infrastructure and operations. That intersection is exactly what makes the project interesting to me.
Build the whole loop
I am not approaching Mon Réceptionniste as a single AI feature. It is a system that begins before the call and continues after it.
Before the call, the product needs the right business context and clear boundaries. During the call, it needs a fast, understandable conversation. Afterwards, it needs to turn what happened into something useful: a summary, a notification, an appointment or a next action.
I am working directly across much of that loop — from product and interface decisions to voice infrastructure, messaging, analytics and distribution. Coding agents make it possible for me to build more of the system myself, but they do not decide what should exist or whether anybody will use it.
Distribution is part of the product
Building software has become faster. Reaching the right users has not.
Professionals do not wake up wanting a voice agent. They want fewer interruptions, more captured opportunities and a service that stays responsive. The way the product is described, demonstrated, onboarded and trusted matters as much as the underlying models.
That means distribution is not a task that starts after the product is finished. It shapes what gets built. A promise that is difficult to explain is often difficult to deliver. A use case that cannot be demonstrated clearly may not be focused enough yet.
Mon Réceptionniste is therefore both a product and an ongoing distribution problem. I want to learn how the two reinforce each other.
Why write about it
I do not want to publish a polished retrospective after every difficult decision has been edited out.
I would rather document specific work while it is still useful: adding messaging after calls, designing agent boundaries, testing acquisition channels, changing positioning and understanding where the system fails.
This site is where those notes will live. Not as proof that I have solved voice AI or SaaS distribution, but as a record of what I am actually building and learning.