DataPal is building the governance layer that AI agents need to operate autonomously without compromising user privacy. Most current agent frameworks struggle with the permission problem—how to give an agent enough data to be useful without giving it a blank check to a user's digital life. DataPal's MyTerms framework solves this by creating a set of machine-readable rules that an agent must follow to access specific data points.
Within the agent stack, DataPal occupies the infrastructure and identity layer. They are active in the push for permissioned intelligent agents, moving toward an ecosystem where agents are not just tools for platforms, but extensions of the individual's own digital agency. Their upcoming AI Agent Store suggests a future where users can discover and deploy agents that are pre-configured to respect standardized data sharing protocols.
DataPal builds infrastructure for the exchange of personal data between individuals and organizations. Based in Glasgow at Barclays Eagle Labs, the company addresses the fundamental friction of the modern internet: data is trapped in silos or duplicated across thousands of databases. This fragmentation makes it difficult for individuals to control their information and expensive for companies to maintain compliance with privacy regulations. DataPal proposes a shift from this platform-centric model toward a person-centric architecture where the individual maintains a single, verifiable record of their information and grants temporary access to services.
The technical foundation consists of two primary components: MyTerms and MyKey. MyTerms is a framework for machine-readable consent. While traditional terms of service are written for lawyers and ignored by users, MyTerms creates a set of digital rules that AI agents and organizational systems parse automatically. This allows a user to define exactly how their data is used, for how long, and under what conditions. MyKey provides the portable digital identity required to authenticate these permissions across different services. Together, these tools form a "Recorded Delivery Network" for smart data and provide a trail of accountability for every data interaction.
Co-founder Bart Smuga and the team position DataPal as a necessary evolution for the age of AI. Large language models and autonomous agents require high-quality, personal data to be useful, but current methods of data collection are often opaque or non-compliant. DataPal provides a structured way for agents to request permission, which reduces the risk for companies that want to build intelligent services without the liability of hoarding unverified user data. The company has gained traction in regulated sectors such as financial services and healthcare, where data accuracy and audit trails are mandatory.
Practical applications are visible in the company’s pilot projects. In retail, DataPal enables loyalty programs that operate on user terms rather than store-specific databases. In healthcare, it allows individuals to share wellness data with insurance providers or health apps in exchange for rewards. The user does not surrender ownership of the underlying records. The company also maintains a presence in the Salesforce ecosystem, using integrations to help businesses manage customer data flows more effectively.
As AI agents become a primary way users interact with the web, the demand for a standard way to handle permissions will increase. DataPal is the middle layer that makes these interactions possible. They are currently developing an AI Agent Store, which is a marketplace for services that agree to operate under these transparent, permissioned rules. This move suggests an ambition to move beyond simple data storage and into the facilitation of the broader agent economy. By focusing on open standards and interoperability, DataPal avoids the walled garden trap that defined the previous generation of data management companies. They do not build a vault to hide data away; they build a gateway to share it with confidence.
Machine-readable personal terms for transparent data relationships.