The fact: This week, Anthropic launched Project Glasswing, a programme giving AWS, Apple, Cisco, Google, JPMorgan Chase and Microsoft early access to Claude Mythos Preview — a model designed to identify security flaws in code. In just a few weeks of internal testing, Mythos uncovered several thousand zero-day vulnerabilities: flaws that were unknown, and therefore unpatched, in mature codebases.

The Codito analysis: Until now, AI assisted human experts. With Mythos, it single-handedly carries out a mission previously reserved for highly specialised teams (security audits, red teaming). And it doesn't do it just a little better — it does it at a scale humans cannot reach: several thousand vulnerabilities identified in a few weeks, where a human team would spot a few dozen. This is no longer a copilot. It is an autonomous actor with a mission. Even though Mythos targets cybersecurity, the pattern generalises: accounting audits, compliance, quality control, technical support, due diligence. Wherever a mission consists of “examining everything to spot what's wrong”, AI now holds a structural advantage.

What it means for you: If you produce code, sell SaaS, or your operations depend on digital tools (few companies escape at least one of the three), ask yourself: which players around you will deploy this type of model within the next twelve months? The dangerous competitors are not the ones with an eighteen-month head start. They are the ones who have started orchestrating agents that, themselves, hold a structural advantage.

The number of the week 30 billion dollars. On 11 May, Anthropic officially confirmed it had crossed this annualised revenue threshold, overtaking OpenAI ($24bn) on the battleground that truly matters — the enterprise. Last week it was a Wall Street rumour; this week it is an accounting reality.

Allow me a different format this week. Rather than a product pitch, here are five questions our prospects are thinking — the ones nobody dares say out loud in a meeting. Our honest answers.

1. “Will AI really replace my teams?”

No. But that's not the right question. The real one: will your AI-equipped competitors produce as much as you do, with fewer people? The answer is yes. What changes: productivity per head. The leaders who succeed don't cut jobs; they reinvest the gains into higher-value work. The ones who fail keep headcount flat and watch their margins erode.

2. “What does it really cost to implement AI in an SME?”

Far less than you think, and far more than the software vendor calling you promises. A structured AI Audit starts at €1,400 excl. VAT with us; a monthly AI Partnership at €349 excl. VAT. The heavy part is almost never the software — it's the time spent on scoping, documentation and change management. Expect two to six months for an AI project to deliver measurable ROI, not two weeks.

3. “Why pay an AI consultancy when ChatGPT is free?”

Because you are paying for the strategy, not the tool. ChatGPT answers questions. A consultancy frames the ones that need asking: which processes to apply AI to, which risks, which metrics, what sequencing, what governance. Without that framing, ChatGPT costs you the time your teams waste — and that time is anything but free.

4. “How long before what you're selling becomes obsolete?”

The tools we would recommend today will have mutated within twelve months — look at Mythos, released this week. That is precisely why our AI Partnership is monthly and cancellable at any time, not a three-year framework contract. We charge for a discipline of constant reassessment, not a frozen technology.

5. “If AI is so good, why haven't you replaced your own consultants with it?”

We have — for a significant share of our work. Monitoring, synthesis, first drafts, initial diagnostics — all of that is now handled by agents we orchestrate. What remains human: listening to the client, making the judgment calls, owning the recommendation. That is exactly what we charge for. Not the execution — the decision.

If one of these questions is the one you've genuinely been asking yourself, tell us. A direct conversation is always possible — no commitment, no pushy salesperson.

Book a conversation

Discover our Codito offers

To close this edition, a word about our back office.

For several months, we have been working internally on building our own CRM and ERP tools — custom-made, designed around our method and our workflows, powered by the AI agents we orchestrate every day for our clients.

Development is now complete.

Over the coming weeks, we will progressively migrate all of our clients onto this new stack. In concrete terms, this means:

• Sharper, faster tracking of every engagement
• A better-structured client channel — history, deliverables and exchanges synchronised in one place
• Greater responsiveness on urgent matters

It's a change that is invisible at first glance — yet it changes everything about the quality of our service. And it is also the most direct demonstration of our method: we apply to Codito what we recommend to our clients. We document, we structure, then we let AI agents execute whatever can be executed.