Articles de Ahmed
-
The SOC as the Enabler of a Self-Healing Posture
The SOC as the Enabler of a Self-Healing Posture
There is a metaphor that security teams use so often it has stopped meaning anything: the alert queue as a backlog to…
6 commentaires -
The Tesla vs. Waymo of Cybersecurity: Why Security Operations Can't Rely on Pure AI
The Tesla vs. Waymo of Cybersecurity: Why Security Operations Can't Rely on Pure AI
In the race to build fully autonomous vehicles, two fundamentally different approaches have emerged: Tesla vs. Waymo.
2 commentaires
Activité
4 k abonnés
-
Ahmed Achchak a partagé ceciAI made the SOC faster. The next step is to make it stronger. That's the evolution I'll be unpacking at Infosecurity Europe on June 2. Alert triage with AI is largely solved. We've seen it ourselves at Qevlar AI: our customers, Fortune Global 500 companies, and leading MSSPs have achieved 90% less manual workload, 9x faster investigations, and real production deployments at scale. But attackers don’t stand still. Finding and exploiting gaps now takes minutes, and the gap between attack capability and SOC capacity is widening. Speed alone doesn't close it. What actually does? Three things: → Context that isn't fragmented. The problem today is that every tool holds its own picture, and no one ever sees a unified view. → Knowledge that compounds: what one analyst learns today shouldn't walk out the door with them. → Teams that stop working in silos: SOC and vulnerability teams prioritizing risk from completely different pictures is no longer acceptable when attackers move in minutes. The session will cover where most teams are today, where they get stuck, and what it takes to build defense that gets stronger with every investigation. 📍 ExCeL London, Theater: AI & Cloud Security 🗓️ June 2, 4:30 PM
-
Ahmed Achchak a publié ceciA couple of weeks ago, I was talking to a senior security leader who'd abandoned an AI investigation tool after a 3-month pilot. I asked him what went wrong. It performed well most of the time. But the moment a conclusion looked questionable, analysts had no way to validate how the system got there. They only got the confidence score and the conclusion. His senior analysts started routing around it within weeks because a system that is “usually right” but impossible to interrogate creates a very specific operational risk. Analysts can’t validate conclusions during high-impact incidents. They can’t explain decisions to leadership or auditors. And they don’t know whether the AI missed context, skipped a step, or hallucinated entirely. I've heard versions of this story multiple times from different prospects. There are many (many) AI SOC vendors on the market right now. None of them demo badly. The difference only appears over time. Because eventually, every SOC team runs into the same question: When your AI gets it wrong, can my analyst see exactly where the reasoning broke down? We took a different path with Qevlar AI: show the work, all of it, even when it's almost too much. → Every step of the reasoning → The exact queries Qevlar AI ran in each of your tools → What our AI liked or disliked in any piece of log it collected → Every severity call, with the explanation behind it That's what it takes to earn trust instead of becoming the black box we're talking about. And because every environment is different, analysts can embed their own context directly into the system. Over time, it stops being an AI and becomes THEIR AI. My DMs are open for anyone who wants to make this reality for their SOC teams.
-
Ahmed Achchak a partagé ceciYour SOC is not supposed to close alerts. That is what it does. It is not what it is for. The job of a SOC is to produce a defensive posture that gets harder to breach over time. Closed alerts are the mechanism. They should also be the input: every verdict, every dismissed false positive, every resolved case is intelligence about your environment that the next investigation should be able to use. Most SOCs are not built that way. They are built around events, not memory. An alert fires, gets processed, closes. What the analyst learned evaporates the moment they log off. The next shift starts from the same baseline as the one before it. The result is a SOC that runs the same playbook on day 1,000 that it ran on day 1. Just faster. The posture does not compound. The attacker's room to maneuver does not narrow. The queue just keeps moving. The real question is not how fast your team closes tickets. It is whether your defense is measurably harder to penetrate today than it was six months ago, as a direct result of every investigation your SOC ran in between. I wrote about what it actually takes to make that true.
-
Ahmed Achchak a publié ceciMost people talk about AI SOC tools as triage tools and measure how fast they close alerts and how much analyst time they save. Those metrics are the first pocket of ROI we are seeing as well. Two or three years ago, when we started, this was still a hard sell. Today, it is an established truth. But this is just scratching the surface of what a good AI SOC tool can do. And now the efficiency that used to deliver a “wow” is table stakes. This means the more interesting conversation can begin. The true value of AI should not be limited to response only. It is what the system retains after the response. Every investigation is a signal. “This entity behaved this way." "This pattern resolved benign forty times." "This attack chain matches something seen 3 months ago in a different part of the environment.” A system that investigates and forgets is just automation. And I am bullish on AI opening the world to a system that investigates and learns. This system is building an indexed, continuously updated map of how attackers move through your specific environment. That map is what makes life harder for attackers over time because it narrows their options. → Techniques that worked before get recognized earlier → Patterns that looked unrelated start connecting → The environment becomes less legible to the adversary and more legible to the defender THIS is the shift most AI SOC conversations miss. When evaluating AI’s performance in your SOC, only one thing really matters: is your defensive posture measurably stronger six months from now than it is today? Because it is a direct consequence of every investigation the system ran in between. Speed is a feature. Indexed knowledge is a compounding advantage. ✅
-
Ahmed Achchak a publié ceciThe era of patch-and-pray is over. Claude Mythos Preview autonomously found and exploited vulnerabilities that had survived 27 years of security audits, bug bounties, and human review. Across every major OS and browser. The time from disclosure to working exploit now collapses to hours. And the real shift goes beyond speed. Attackers now operate at scale. The attack surface got harder to defend because the volume of viable attacks just multiplied. Vulnerability management was built on a different premise. The whole model assumed attackers needed comparable effort to weaponize a known weakness. That assumption is gone. AI can now turn a CVE and a git commit into a working exploit autonomously. For legacy or undefended systems, the window between disclosure and exploitation is already closing faster than most programs can respond. Patch SLAs written for multi-week remediation cycles are already wrong. Detection can't be bottlenecked by analyst queue depth or shift coverage. If exploitation happens at scale and at speed, investigation needs to run the same way: continuously, consistently, on every alert, which is possible with Qevlar AI. It is the baseline for what's coming.
-
Ahmed Achchak a publié ceciEvery security investigation creates knowledge. But most SOCs still investigate too many alerts as if they are seeing them for the first time. That is the real gap. Every investigation produces reusable context: → Relationships between observables → Patterns already validated as benign or confirmed as malicious → Decision paths that led to the right conclusion Analysts absolutely build expertise. But too often, that expertise stays tied to a person, a shift, or a moment in time. Its reuse depends on memory, availability, and time, all of which vary across teams, workloads, and handoffs. That is what creates inconsistency. Based on Qevlar AI's data, around 40% of alerts can be enriched when similar past investigations are surfaced. A meaningful share of investigations already has relevant precedent, but captured knowledge remains passive. For prior knowledge to influence the next investigation in real time, three things need to happen: → Similar cases need to be recognized → The relevant prior reasoning needs to be extracted → That context needs to appear directly inside the analyst workflow Without that, reuse stays incidental. And under pressure, incidental processes do not hold. With Qevlar AI's intelligence layer, this can happen autonomously. Knowledge that used to disappear with the end of a shift, or walk out the door when an analyst leaves, becomes part of how the next investigation runs. That is how expertise compounds. Over time, this gives teams a stronger, more consistent understanding of their security posture.
-
Ahmed Achchak a partagé ceciSIEMs have been promising to surface the full attack picture for two decades. Yet, most post-mortems still end with an analyst manually rebuilding the attack timeline. So what’s the main blocker? I sat down with Raffael Marty, one of the people who actually built the SIEM space (ArcSight, Splunk, ConnectWise, 25+ years in security), and we went deep on this. He nailed the root of the problem right away: the industry keeps building smarter tools on top of fundamentally flawed inputs. The alerts feeding your SIEM were misconfigured to begin with. You don't control how they're generated. You have no idea what you're not seeing. And most AI-native SOC tools today are doing exactly the same thing: adding a reasoning layer on top of that noise and calling it a solution. What we actually need is an intelligence layer that keeps the institutional knowledge that lives in people's heads and walks out the door when they leave. A layer that learns from analysts, encodes that knowledge, builds on it, and applies it automatically to every investigation. So that even incomplete signals can be understood for what they actually mean and help analysts remediate faster. That’s exactly the vision we have at Qevlar AI, and in the episode, I took the time to decode the most challenging questions around it. The full discussion covers: → Why SIEMs drifted from cross-source correlation to single-source noise (and what’s wrong with it) → Campaign-based thinking (a fundamentally new approach), what it looks like day-to-day, and the role of AI in it → Why LLMs alone can't hold the context at scale, and what the only way to stabilize them is And much more. 🎧 Listen to the full episode: https://lnkd.in/e4u9TZ84
-
Ahmed Achchak a partagé ceciThis week, we announced our new funding round. $30M to help SOC teams see the bigger picture across attacks and campaigns. Next up: RSAC 2026, and Qevlar AI's team is getting ready in full force. I’ll be joining AGC Partners' panel of experts from leading AI companies for a discussion, “Building Cyber Leaders in EMEA: Ecosystem Dynamics, AI Disruption & Global Exit Pathways”. The EMEA cybersecurity ecosystem is producing a new generation of companies grounded in deep technical expertise, strong enterprise relationships, and increasingly AI-native architectures. I’m looking forward to exploring how these shifts are redefining security platforms and how AI is being deployed across SOCs in production environments. ⏰ Monday, March 23 | 4:30–5:15 PM 📍 Floor 9, Grand Salon, The City Club of San Francisco (155 Sansome St, San Francisco, CA 94104) There will also be a private cocktail afterwards to continue the conversation. For any inquiries regarding the event, please contact Meghana Reddy at mreddy@agcpartners.com.
-
Ahmed Achchak a partagé ceciFor as long as I can remember, I've been describing Qevlar AI to our team with the same metaphor: an airplane. In the early days: engines on, vision locked, and building speed to lift off. Today, Qevlar AI is at full altitude with global enterprises and leading MSSPs running their investigations on our platform. Now it’s time for the next stage. I’m proud to announce that Qevlar AI has raised $30M in a new funding round backed by Partech, Forgepoint Capital, and EQT Ventures & Growth. The funding is fueling a clear mission. SOC teams are at their best when they can think at the level of patterns, campaigns, and impact instead of individual alerts. That kind of visibility has always existed in theory. We’re determined to make it a reality. We're building the platform that surfaces what matters, eliminates the noise, and gives analysts the time and clarity to focus on real threats and response. To the Qevlar AI team, our customers, and everyone who believed early — thank you. To every cybersecurity leader who is building a top-performing SOC — we should talk. Our next event is RSAC, booth S-0254, March 23-26. See you there! And if you’re in Europe, see you at Forum INCYBER (FIC), booth D13, March 31-April 2! @Hamza Sayah, Hakim Jakhjoukh , Fraser Whitfield
Ahmed Achchak a commenté un post
3 sem.
Welcome to the team, Sam Wilson!
Ahmed Achchak a répondu à un commentaire
1 mois
Thanks Philippe Alegre, the feeling is mutual. Excited about what we're building together, and this is only the start!
Ahmed Achchak a répondu à un commentaire
1 mois
Rob van Os It was a pleasure having you on the podcast!
-
Ahmed Achchak a aimé ceciAhmed Achchak a aimé ceciDetection and response that runs on periodic, batched logs worked when attackers moved at human speed. Against agentic AI attackers, it's a structural liability. In the latest #ForgepointTIPS, Dr. Shane Shook breaks down why detection and response needs to become continuous and event-driven to operate at the speed of AI-enabled attacks. Featuring perspectives from Derek Collison, Nanda Santhana, and Ahmed Achchak. Read more 👇 #agenticAI #cybersecurity #forgepointfamilyTIPS #41: The Clock Starts Before You Think It DoesTIPS #41: The Clock Starts Before You Think It Does
-
Ahmed Achchak a aimé ceciSuper heureux de pouvoir poursuivre cette initiative avec mes "partners in crime" 😉Ahmed Achchak a aimé ceciTrois boîtes cyber françaises sur un terrain de padel. Ça ressemble au début d'une blague. C'est une league. On organise avec StrangeBee une padel league privée, et EVERTRUST s'y associe. Au padel, vous pouvez avoir le meilleur coup droit du terrain : si votre partenaire ne couvre pas sa zone, vous perdez le point. Personne ne gagne seul. La cyber, c'est exactement pareil. La détection et la réponse à incident chez StrangeBee. La confiance numérique et les certificats chez EverTrust. La gouvernance, le risque et la conformité de notre côté. Trois zones du même terrain. Trois métiers qui se parlent rarement, alors qu'ils défendent les mêmes clients. Si vous êtes un(e) professionnel(le) de la cyber vous pouvez demander à intégrer la league : https://lnkd.in/ejdxhD_P Les places sont limitées, mais pas de panique ! Si nous sommes complets pour cette édition et que votre demande est validée, vous serez prioritaire pour nos prochains événements. Et chez vous : la sécu opérationnelle et la GRC, elles se parlent tous les combien ?
-
Ahmed Achchak a réagi à ceciAhmed Achchak a réagi à ceciThis weekend I attended my father's graduation from the University of Cambridge. Usually it's the parents tearing up at a graduation. This weekend, the roles were reversed. Class of '04. And how proud I am that, 22 years later, he could finally walk at his ceremony, with both my sister and me there to see it. My father left Turkmenistan as a young man with little more than ambition and the belief that if he worked hard enough, a greater life was possible. Europe gave him what every young person chasing a dream deserves: a chance. A chance to earn a world-class education and to prove himself, to fulfil the ambitions he'd carried with him, the foundation that led to a international career at the UNDP. There's a certain hunger that comes with being first generation, the drive of people who've had to earn and build everything from the ground up. It's also the mindset I recognise every day in my team at Qevlar AI, built in France, with a deeply international team that knows what it means to build from scratch. Congratulations, Papa!
-
Ahmed Achchak a réagi à ceciAhmed Achchak a réagi à ceci📈 +10.6 % C’est la progression de notre chiffre d'affaires à 2 chiffres sur ce premier semestre 2026, avec 3 300 experts engagés autour du monde et face à une menace qui ne dort jamais ! 🇪🇸 Dans le cadre de notre stratégie de croissance et des ambitions du plan #TrustTheFuture, nous avons lancé, en mai 2026, une nouvelle filiale en Espagne (⚽⭐⭐), Orange Cyberdefense España en collaboration avec MasOrange, sur le 5ème marché cyber en Europe, renforçant ainsi notre empreinte géographique. Nous intensifions également l’intégration de l’intelligence artificielle grâce à des partenariats stratégiques avec des fournisseurs de plateformes d'IA comme Qevlar AI 🤖 🤝 Ensemble, et après nos 10 années d’expertise et d’innovation, continuons de construire une société numérique plus sûre. Orange | Orange Business | Hugues Foulon | Christel Heydemann | Laurent Martinez | Vincent Lecerf | Aliette Mousnier-Lompré | Adela Guédon| Paz Fonteboa | Olivier de Paillerets | Frederic Zink | Kaja Narum | Geoffroy Sirven-Vienot #SaferDigitalSociety
-
Ahmed Achchak a réagi à ceciAhmed Achchak a réagi à ceciOur Venture team has already got a lot under their belt this year! ⚡ The Venture strategy backs European companies at Series A and Series B and provides hands-on support in building the next generation of category-defining technology. Since January, these are just a few notable milestones from the portfolio: - ABTasty joined forces with VWO and exited our portfolio - Qevlar AI joined the portfolio with $30M in funding - Smartness raised a $47M Series B round - Fresha reached unicorn status And on the fund side: - We picked up a major industry award - Simone Riva Violetta appeared on the Startuprad.io podcast and an article in Technology Magazine - Reza Malekzadeh joined Ahmed Achchak, Qevlar AI’s cofounder, on the EUVC podcast to discuss modern cybersecurity The first half of the year brought a new company, a record-breaking round, a unicorn, and lots more. We’re not done yet though, and H2 promises just as much momentum! Find out more about each of these achievements in the comments below! CC: Vivien Bovard, Philippe Collombel, Reza Malekzadeh, Julia Najman, Magda Posluszny, Simone Riva Violetta
-
Ahmed Achchak a réagi à ceciAhmed Achchak a réagi à ceci🚀 I'm thrilled to share that Forgepoint Capital International is co-leading Multiverse Computing’s €500M ($570M) Series C funding round alongside BNP Paribas, and Bullhound Capital with participation from a consortium of global investors including Santander, Tikehau Capital, HP, Orange, SETT Spain and EIC - European Innovation Council. When we first invested in Multiverse Computing last year, we were confident they had the potential to disrupt AI inference economics with optimized LLMs that drastically reduce compute costs. Our conviction has only deepened since. Multiverse has grown their product into a complete AI foundry and operating system trusted by leading global organizations across manufacturing, finance, energy, aerospace, cybersecurity, defense, and health and life sciences. Today, they address a much larger problem: deploying efficient and secure AI at scale, with observability, security, governance, multi-tenancy, and chargeback built in across the cloud, at the edge, and in data centers. The decision to double down on our investment was simple. Multiverse is the only company we’ve seen with the technology, team, and traction to become the critical efficient and secure AI platform for the next phase of AI deployment at scale, and they’re just getting started. Congratulations to Enrique Lizaso Olmos, Román Orús, Samuel Mugel, PhD, and the Multiverse Computing team! We look forward to continuing our collaboration. 💡 Read why we doubled down on Multiverse Computing in the comments 👇 cc: Marta Garcia Pedruelo, Harshit Krishna, Leire Herrero, Asier Arburua, Alberto Reguera, Jaime Goyarrola #forgepointfamily #AI #Deeptech #cybersecurity #DefenceTech #EU #Sovereignty
Expérience et formation
-
Qevlar AI
Voir toute l’expérience de Ahmed
Découvrez son poste, son ancienneté et plus encore.
ou
Prix et distinctions
-
Winner - Huawei Big Data Challenge in France 2018
-
Ranked 1st over 40 teams from 10 top schools and universities in France - Prize: 10 000€.
24-hour Machine Learning challenge - Developed an innovative NLP-inspired algorithm for Malware Detection as part of a 4-person team and presented the winning solution in front of the jury - AUC > 0.99 (Python). -
Moroccan Merit Scholarship
Ministry of Higher Education and Research
Awarded to top ranking students in competitive entrance exams to French Grandes Ecoles
Langues
-
Arabic
Bilingue ou langue natale
-
French
Bilingue ou langue natale
-
English
Capacité professionnelle complète
-
Spanish
Notions