Askeal takes the opposite approach to omniscient Gen AI: rather than pretending to know everything, it combines AI with community expertise. Vetted vendors, researchers, and practitioners contribute their intelligence and tools to help users conduct manual investigations. The startup, backed by a $1.1 million pre-seed round, is launching the tool with an international community of early testers and contributors.
A SOC analyst can start the day to 50 alerts requiring manual review. An IT manager may face a similar challenge while managing multiple tools and systems. They interact with Askeal in natural language and attach the files or other evidence they need. Askeal analyses the information, checks it against relevant sources, and delivers evidence-backed assessments with verifiable sources. That order of operations is the product.
Roxane Suau leads Askeal, a company she cofounded in August 2025 that closed its first funding round three months later. She describes the solution in terms that leave the technology in second place. “It’s an AI cybersecurity assistant powered by community expertise, but we often drop the AI part because the community is what makes it unique,” she says. What she puts first is Askeal’s network of expert contributors whose knowledge the assistant draws on, and the transparency that lets users see the raw intelligence behind every answer.
What comes back when you ask a question
Ask about a specific CVE and Askeal gathers related technical research, published proof-of-concepts and available patches, then provides a risk level and recommended actions. You come across a suspicious URL, paste it into Askeal, and get the verdict of vendors’ tools, a check of phishing databases, and intelligence on the campaign it is associated with.
Askeal gathers knowledge from private and open-source feeds. Where available sources are insufficient, Askeal leaves the relevant section empty rather than filling it with weak or irrelevant material. “We are not afraid of letting a section empty if there is nothing relevant to share,” Suau says.
Beyond question answering, the cybersecurity assistant supports investigative tasks such as log analysis, threat detection through integrated contributor capabilities, URL, domain and hash lookups, CVE remediation guidance, and security configuration recommendations. A user can also upload a log file containing several indicators of compromise, Askeal extracts the assets, then checks each against its available sources.
270+ vetted contributors, amplified by Askeal’s technology
Every contributor is approved manually before joining. The team vets them against criteria for expert-level knowledge. More than 90 contributors have come on board since the program started around two months ago, joining the 178 public sources that were initially added to the tool.
After the human check, layered AI review scores the quality and the technical level of submitted content, and when a user asks a question, the system corroborates answers against one another. Contradictions survive. If one source calls something safe and another calls it dangerous, both appear, because a disagreement between two credible experts is information an analyst can use. Users can rate the quality of answers, providing a feedback signal alongside source verification.
The company has developed its neuro-symbolic technology with cybersecurity and AI researchers in collaboration with the Montpellier Laboratory of Computer Science, Robotics, and Microelectronics. It is now bringing it to market while continuing to evolve how generative AI can support cybersecurity operations.
It sits next to the tools you already own
Most target customers already run an EDR or XDR with AI inside it, and Suau does not ask them to switch. Her argument is about the size of the pool each system draws from and the work that remains once an alert has been raised. “The AI will just take a decision based on the amount of intelligence they have within that tool,” she says of the built-in assistants, which answer from what a single vendor knows. Askeal draws on what the wider ecosystem knows, then shows the intelligence to the person making the decision.
There is a second argument, aimed at a claim that circulates in the market. Vendors selling AI-driven SOC automation talk about closing 70 to 85 percent of tickets without a human.
Practitioner forums do not describe anything close to that. Manual investigation has not gone anywhere. At some managed security providers a single analyst still works through 50 or more alerts a day, and each of those alerts needs a person to gather context and decide. That gap is the market. In the long run, Suau sees Askeal feeding enrichment into automated SOC platforms. For now it works directly for the person doing the investigating.
A community that grows fast
Askeal opened its beta in February 2026 with a target of onboarding 300 testers in 3 months and reached 500 in two and a half months only. The team recruited voluntary testers through direct outreach and has used their feedback to build the product, including deciding which features to prioritize and what to improve.
SOC analysts represent roughly half of the early community. Other users include IT managers, security engineers, and administrators who handle security work without specializing in it. The early community spans 69 countries and 14 roles across IT and security. Commercially, Askeal is primarily focused on medium-sized companies and managed security providers, where teams often need more capacity or specialized expertise.
Freemium today, as the product moves toward its official launch
The first round of $1.1 million funds the push to market and the international scaling that follows it. Askeal is free today. Paid plans are coming, and with them a revenue share that pays contributors a percentage based on how often their expertise is used in answers.
Anyone can put a question to Askeal without an account, though the answer is trimmed. A free account gets considerably more, with free daily credits. The free tier will remain alongside the paid plans in reduced form, which makes the present a good moment to hand it a domain, a log file, or a CVE and see whether the sources hold up.
