Pomegra
London startup CodeWall exits stealth with $4M pre-seed led by Osney Capital to sell autonomous AI attack agents as a continuous alternative to annual pentests.
Key Takeaways
- CodeWall raised a $4M pre-seed round led by Osney Capital, announced October 14, 2026.
- InReach Ventures, Antler and angel investors joined; valuation was not disclosed.
- Its AI agents attack enterprises from outside with no credentials, aiming to replace yearly penetration tests.
Lead
CodeWall, a London-based offensive security startup, emerged from stealth on October 14, 2026 with $4 million in pre-seed funding. Osney Capital led the round, with InReach Ventures, Antler and a group of angel investors participating. The company, founded and run by CEO Paul Price, builds autonomous agents that probe a company's external attack surface the way a hostile party would, then document each path that works.
The round values the business at an undisclosed figure. CodeWall said it already works with banks and financial institutions.
What Does CodeWall Actually Sell?
CodeWall sells software that runs penetration tests continuously instead of once or twice a year. Its agents start with no credentials, no source code and no internal access, which means they see what an outside attacker sees. Several specialised agents work on an assignment at the same time and write their findings to a shared investigation graph, which produces a step-by-step record of how each weakness was reached.
That pitch targets the traditional pentest, a point-in-time engagement by a consultancy that is out of date soon after the report lands. It also lands in a crowded field. Automated pentesting and exposure-management vendors have multiplied, and many promise the same continuous coverage. CodeWall's differentiation rests on the depth of its attack chains and the evidence it attaches to them.
Why Did Three Consulting Giants Matter to This Launch?
The firms matter because CodeWall used them as public proof of its agents' reach. Earlier in 2026 the company's agents found vulnerabilities in platforms belonging to McKinsey & Company, Boston Consulting Group and Bain & Company. Against McKinsey's Lilli AI platform, an agent gained read and write access to the production database within two hours, with zero credentials.
The agents also surfaced an exposed database at BCG and an exposed competitive intelligence platform at Bain. According to the company, all three firms fixed the problems within days, and McKinsey's forensic review found no unauthorised access to data. Those results are self-reported by CodeWall, but the speed of the remediation suggests the findings were taken seriously.
How Will the $4M Be Spent?
Most of the money goes toward widening the range of environments the agents can investigate. CodeWall also plans to build verification and remediation workflows, so a finding can be confirmed as exploitable and later confirmed as closed. Hiring will cover offensive-security, engineering, customer-success and go-to-market roles in London.
A pre-seed of this size buys roughly a year of runway for a small technical team. The investor mix is telling: Osney Capital, InReach and Antler all back early European companies, and none is a late-stage cybersecurity specialist. The round says little about CodeWall's market share, and more about investor willingness to fund AI-native security tooling before revenue is proven.
What Comes Next for AI-Driven Offensive Security?
The next test is conversion from headline research to repeatable contracts. Naming three prominent consultancies generates attention, but enterprise buyers will ask about scope control, false-positive rates and the safety of letting autonomous software attack production systems. Banks, CodeWall's stated early customers, will press hardest on all three.
Competitors with larger balance sheets can copy the continuous-testing message quickly. CodeWall's defence is the quality of its evidence trail and how fast it moves from finding to verified fix. If the remediation workflows ship on schedule, the company can argue it sells outcomes rather than reports. If they slip, it risks being filed alongside the scanners it set out to replace.
Outlook
CodeWall enters the market with a $4M cushion, a named lead investor and a set of high-profile findings. Valuation and revenue remain undisclosed, so the round is best read as a bet on the team and the approach. The next 12 months of customer wins and remediation features will show whether continuous AI-driven attack testing becomes a standard budget line or stays a feature inside larger security platforms.
