The Atlas Project

1. How it all started

The story of Atlas doesn't start in 2026. It starts much earlier, in the accumulated frustration of people who work with software every day and feel like something fundamental is broken.

For Thibault, the first signal came in 2021, during the Covid years, when he launched Tribes, a French-language media project designed to let professionals share honest, operational feedback on how they actually run their companies. What started as a broad experiment quickly revealed something unexpected: an enormous amount of the conversation was about software. How to choose it, how to implement it, what worked, what didn't, and what nobody tells you before you sign the contract.

At the same time, Thibault had spent ten years building Partoo, growing it to €40M ARR and a team of over 300 people. Partoo works closely with Google Maps and manages more than 350,000 Google Business profiles on behalf of large companies. That experience gave him a deep understanding of something that would later become central to Atlas: the relationship between trust, reviews, and business reputation. He also spent years selling software to large enterprises, watching procurement teams make decisions based on feature lists and slide decks, rarely on the real experience of people who had actually used the product.

For Alexandre, the signal came from the other side of the table. After years of consulting and then building at Murfy, a French startup, he had lived through the chaos of choosing tools for a scaling company. Alexandre did what most managers end up doing in that situation: he texted peers in similar roles at other companies, asked what they were actually using, what they'd dropped, what they wished they'd known. Those informal conversations were systematically more useful than weeks of vendor research.

We had been flatmates for years, and the conversation kept coming back to the same frustration: why is it still so hard to make a good software decision?

And eventually, we decided to do something about it.

2. The world we see

Software is eating the world. This is not a new idea, but it is one that keeps accelerating in ways that most people haven't fully understood yet.

For years, the number of software products available to businesses has grown steadily. Today, that growth is entering a new phase. Vibe-coding, the ability to build software through AI-assisted development, means that almost anyone can ship a product. The barrier to creating a SaaS has never been lower. And as a result, the number of tools available to businesses is going to keep growing, faster than ever before.

We think this is fundamentally a good thing. More software means more innovation and more opportunities for companies to find tools that genuinely fit the way they work.

But abundance creates its own problems.

When there were 50 project management tools on the market, choosing one was manageable. When there are 500, it gets impossible. How do you know what actually works? How do you keep up with a landscape that evolves faster than any procurement team can track?

We also believe something deeper about how companies work. A company is no longer just the sum of its people. It is the sum of its people, its software stack, and increasingly, its AI agents. The tools and agents you choose shape how your teams communicate, how decisions are made, and how fast you can move.

The stack is no longer a technical detail. It is a strategic decision, and it deserves to be treated as one.

3. What about existing software review platforms?

The platforms that dominate software discovery today were built fifteen years ago, in a different internet, for a different world. At the time, they revolutionised the way businesses discovered and evaluated software, bringing transparency to an ecosystem that had very little of it. They built something valuable, and the software industry owes a lot to what they created.

However, most of these platforms were designed with vendors in mind, not buyers. Their business model depends on software companies paying for visibility, badges, and placement. The result is a system where the loudest voices are not the most honest ones, but the most incentivised. Reviews are requested by sales teams, triggered by NPS surveys, and sometimes rewarded with gift cards. The data looks impressive but the signal is partially biased.

What we find exciting is everything that remains to be built.

When you are trying to decide whether to adopt a new tool, what you really want to know is what people like you are using, and what they think of it. Professionals in similar roles, similar industries, similar company sizes. We believe that the social dimension of software discovery is one of the most interesting problems left to solve. And that contributing to a platform should feel rewarding, that sharing your expertise should be recognised and build into something meaningful over time.

We think there is something worth building here.

4. Our vision for Atlas

Atlas starts with a simple belief: that the best people to help you choose a software or an AI tool are the ones who have already used it, in companies like yours, facing problems like yours. Everything we build flows from that idea.

The stack is at the heart of Atlas. When a professional joins and declares the tools they use, they are contributing to a living map of how software is actually used in the real world. Over time, these stacks become one of the most valuable things on Atlas: a real picture of what works together, what gets replaced, and how technology choices evolve as companies grow.

Alongside the stack, we are building a contribution experience that feels worth coming back to. Sharing expertise should be recognised. The more you contribute, the more you unlock access to high-quality, actionable insights from people who have genuinely been through it. Atlas is designed so that the depth of what you get out of the platform grows with the depth of what you put in.

The social layer is where things get interesting. When you are evaluating a new tool, the most valuable signal is not a star rating from a stranger. It is knowing that someone in your network, a former colleague, a peer in a similar role, has used it and has an opinion. Atlas will help surface those connections, to let you see what people like you are actually using, and to make it easy to reach out and have a real conversation.

But we are also thinking further ahead. The way companies buy software today is broken in a way that goes beyond reviews. Most procurement processes start from the wrong place: companies write requirements based on what vendors tell them exists, rather than starting from the problem they are actually trying to solve. The result is RFPs that favour the loudest vendors, not the most relevant ones.

We think AI can change this. Over time, we want Atlas to offer a conversational layer that lets companies describe their situation, their constraints, and their goals, and receive recommendations built from real ecosystem data and verified user experiences. An agent that can analyse an existing stack, identify gaps, and suggest combinations of tools that actually work together. A process where the starting point is the problem, and where no single vendor gets to define what the answer looks like.

5. Trust in the age of AI

When a company adopts a new software tool, it is making a bet. A bet that the product will hold up in production, that it will handle the edge cases, that it will not break at the worst possible moment. That kind of trust is hard to earn and easy to lose.

Moreover, comparing features on paper is often misleading. Two products can both claim to offer analytics, and have almost nothing in common in practice. The only way to understand the difference is to talk to people who have used both, in real contexts, with real data.

The same logic applies to reliability, to support quality, to the experience of actually implementing a tool and keeping it running. These things do not show up in a product brochure or a slide deck. They show up in the honest feedback of people who have been through it.

And then there is the security dimension, which is becoming harder to ignore. As vibe-coding lowers the barrier to shipping software, it also lowers the bar for what gets shipped. Security architecture, data handling, access controls, these require experience and rigour that fast-moving development cycles do not always allow for. At the same time, AI is making attackers more capable and more efficient. Organisations that introduce a poorly secured tool into their environment are not just taking a product risk. The threat is real, and it will grow.

In this environment, TRUST becomes the most valuable thing a software company can have. Certifications like SOC2 or ISO 27001 are signals worth paying attention to. So is a genuine user base, built over time, made up of people who have run the product in real conditions and are willing to say so. When someone in your network has used a tool and can tell you honestly what they found, that conversation is worth more than any product page, any analyst report, or any sales pitch.

The more software there is to choose from, the more that kind of signal matters.

6. Where we're going

Atlas is an ambitious project. We believe it has the potential to change the way software and AI agents are discovered, evaluated, and trusted, and we are only at the beginning of what we want to build.

Atlas was started by two people, with no external funding and no marketing budget. What you see today is an MVP, made possible by everything that AI and modern development tools now allow, which means that a small team can move faster than ever and build things that would have taken years not long ago. We are genuinely excited about what that makes possible.

That also means that what Atlas becomes depends enormously on the people who use it.

Community projects live or die by the spirit of the people who participate in them. What made Waze work was the millions of drivers who reported an accident, flagged a speed trap, or corrected a wrong turn, day after day, because they believed that the map would be better for everyone if they did. Atlas works the same way. The more people contribute honestly, the more valuable it becomes for everyone.

We are asking people to play the long game with us. To write real reviews, share genuine experiences, and bring the same good faith to Atlas that they would want others to bring. In a world with more software and AI agents than anyone can evaluate alone, an honest review from someone you trust is genuinely useful. We think that is worth protecting.

The vision is still being refined. There are features we have not built yet, directions we are still thinking through, and problems we have not fully solved. We are open to collaborations, to conversations, and to people who want to help shape what Atlas becomes. If you work in software, buy software, or advise companies on software, there is probably a place for you in this.

We also have no budget to spread the word. If Atlas feels useful to you, the most valuable thing you can do is talk about it.

Concretely, here is what makes a real difference:

• Write a detailed review of a tool you know well. A rich, honest review from someone with real experience is worth more than a hundred one-liners.
• Share Atlas with a colleague, a peer, or anyone who has ever struggled to choose the right software.
• If you work at a software company, invite your clients to leave a review of your product (without incentives). Honest feedback from real customers is the foundation of everything we are building.

We are at the beginning of something. We hope you will be part of it.