Company intelligence
NakedAI
AI clarity, architecture, and governance — for organisations that need to get this right.
About NakedAI
NakedAI works with organisations at the point where artificial intelligence stops being theoretical and starts to carry real weight. This is usually the moment when AI decisions begin to affect governance, accountability, risk, or reputation — and when leaders are being asked to move forward without a shared understanding of what is actually being decided. Our role is to bring clarity at that point. We support boards and senior leadership teams to properly frame AI decisions before they are embedded into systems, operating models, or strategy. That means slowing the conversation down just enough to identify intent, assumptions, constraints, and consequences — and to ensure responsibility remains visible rather than dispersed. NakedAI is not a delivery partner and does not build or implement AI systems. We are not aligned to vendors, platforms, or tools. Our independence allows us to focus on judgement rather than momentum, and on decision quality rather than technical novelty. The work typically centres on three connected areas: clarifying what AI is being asked to do in a specific organisational context; shaping the architectural and governance thinking that must exist for that work to be controllable; and establishing clear lines of accountability so decisions can be explained, defended, and overseen over time. Engagements are deliberately focused. They are designed to stabilise thinking, challenge assumptions, and produce clear decision outputs that leaders can stand behind — whether that leads to action, delay, or a decision not to proceed. NakedAI exists because some AI decisions need judgement, not acceleration.
Verified activity
Signals from NakedAI
11 published signals
Research & Knowledge
NakedAI noted that McDonald's did not conclude that the underlying idea of AI voice ordering was dead, as its 2025 annual report shows it was again testing AI voice ordering in selected restaurants on the restaurant technology platform it has been developing with Google.
Reported by Malcolm Scott Maxwell
Research & Knowledge
NakedAI published a study by METR that found that the Human Pause is where the output disappears, moves, or creates complexity.
Reported by Malcolm Scott Maxwell
Research & Knowledge
NakedAI published a study by METR that found roughly half of AI-generated pull requests that had already passed the SWE-bench automated grader would not have been merged.
Reported by Malcolm Scott Maxwell
Research & Knowledge
NakedAI published a study by METR that found failures in core functionality, breakages elsewhere and problems with code quality in AI-generated pull requests.
Reported by Malcolm Scott Maxwell
Research & Knowledge
NakedAI published a study by METR that found developers are making progress faster by early 2026.
Reported by Malcolm Scott Maxwell
Products & Services
NakedAI launched The Human Pause to help answer questions about AI governance and operational models before an impressive number finds its way into a business case.
Reported by Marla Ubhi
Legal & Regulatory
NakedAI noted that in 2025, the Ministry of Justice cited Horizon when it launched a review of whether the presumption that computers operate properly on a system with a lack of transparency remains appropriate.
Reported by Malcolm Scott Maxwell
Products & Services
NakedAI launched The Human Pause to prevent AI-related failures by ensuring people can question and challenge AI decisions.
Reported by Marla Ubhi
Research & Knowledge
NakedAI published an investigation by Canadian Business revealing that Target Canada's product data was initially estimated at 16% accuracy compared to 54-55% in Target's US operation.
Reported by Malcolm Scott Maxwell
Products & Services
NakedAI highlights that Amazon abandoned an ambitious AI recruitment ranking system but retained a more impressive version of the final test.
Reported by Malcolm Scott Maxwell
Research & Knowledge
NakedAI researcher Malcolm Scott Maxwell researched the court case involving fabricated AI-generated legal opinions and human review.
Reported by Malcolm Scott Maxwell