AU & NZ
SG
MY
US
IE
Compliance in the Age of Agentic AI: Access's View from the Front Line
Few sectors carry higher stakes for getting AI governance right than social care. As regulators and sector bodies grapple with how AI should be governed, the risk of moving too fast, or too carelessly, is measured in real consequences for real people. It's a tension Access has approached head-on, embedding agentic AI directly into its platform with compliance and safety built in from the ground up, not bolted on after the fact. Martin Lowthian, Compliance Lead at The Access Group, explains how he sees AI, and agentic AI, reshaping the sector, and why Access is well placed to lead responsibly.
Q: AI is moving quickly in social care, but compliance and regulation often move more slowly. How do you see that gap playing out, and where does Access sit within it?
A: Access is both a thought and solution leader in Ethic AI in software development, strengthened by our certification to ISO42001 and partnership with Microsoft, who have co-presented at Care sector events with us to educate and reassure care providers.
Q: What's fundamentally different, from a compliance perspective, between traditional AI tools and agentic AI that can act on a care provider's behalf?
A: Typically AI tools could allow a user to reactively query their data, the LLM could consume large volumes of data to give an answer; however, agentic AI can be programmed to do a lot of the work that a human did manually and learn what to do and facilitate automated actions, giving time back to the care provider. It flips the 80:20 rule, so you truly have 80% of the time to act, rather than research, query and 20% to take action.
Q: What are the biggest risks the sector needs to get right as agentic AI becomes more autonomous, and how has that shaped Access's approach to building it?
A: The key thing is that care providers are people businesses, so AI and especially Agentic AI must give more time back for human connections. The person makes decisions, reviews what AI is suggesting and makes sure the supported person is always at the centre, with their voice, choice and control.
Q: Trust and accountability are recurring concerns when AI takes on more independent decision-making. How does Access ensure agentic AI remains safe, transparent and auditable?
A: As above, it is about control and ultimate decision-making being done by a human. AI can do a lot of the ‘leg-work’, reducing admin, automating certain elements of a process, but it does not replace entirely an expert; it makes them more productive. If we liken this to when Care staff were allowed to administer medication in a Nursing Home, it did not negate the need for a nurse, it gave them more time to triage and prioritise which residents to visit, spend time with and take a more holistic approach to everyone in the service, rather than a more activity focussed medical model. This made everyone roles more impactful and outcomes better.
Q: How do you expect regulation and best practice around AI in care to evolve over the next few years, and how is Access preparing for that now rather than reacting later?
A: We are actively trying to engage with regulators, including the DPC in Ireland, so we can both understand and endeavour to influence. Care regulators are using AI and we are expecting guidance to be published. It is unlikely to be prescriptive, but we are here to advise and support our customers.
Q: Why does being compliance-led, rather than compliance-constrained, give Access an advantage in leading the sector's adoption of agentic AI?
A: Our approach is to build compliance in at the design stage of any project or solution, so it is never an afterthought or a constraint that holds us back. It is about ensuring we design in compliance to meet the highest standards as a software provider, so care providers and others can meet their legal obligations too.