Agentic automation: A critical enabler for solving the NHS productivity challenge

As the NHS grapples with unprecedented demand, workforce pressures and growing elective backlogs, agentic automation is emerging as the key practical lever of choice

As of September 2025, the median waiting time for patients awaiting treatment stands at 13.4 weeks; a significant increase from the pre-COVID median of 8.0 weeks in August 2019. Meanwhile, 1.61 million people have waited more than four hours in A&E over the past year. The numbers tell a stark story. Behind these statistics lie workforce pressures that compound at every level.

When staff shortages meet rising demand, the administrative burden becomes unsustainable. This is where agentic automation – combining Artificial Intelligence, analytics, and orchestration – offers a strategic response that goes beyond short-term fixes.

The evolution: from simple tasks to intelligent agents

Understanding the automation journey is critical to unlocking its full potential. The NHS’s digital transformation has progressed through distinct phases, each building upon the last.

Deterministic automation marked the starting point. Robotic process automation (RPA) tackles repetitive, rule-based tasks, such as copying data between systems, generating standard letters, and updating patient records. This delivered immediate value, reducing manual errors and accelerating routine workflows, but remained limited to predictable, structured, and linear processes.

The next phase introduced AI-enhanced automation, where Machine Learning and AI capabilities augmented deterministic processes. Natural language processing extracted key information from unstructured clinical notes, and AI deciphered scanned documents, regardless of format. The advent of generative AI enables the rapid summarisation of lengthy patient medical histories.

Now, we’re entering the era of agentic automation – where AI agents operate with genuine autonomy within defined parameters. These agents don’t simply execute tasks; they make decisions, orchestrate multiple steps, interact with humans when needed, and manage entire end-to-end processes.

This evolution matters because NHS productivity challenges exist at the process level, not the task level. Automating individual steps provides marginal gains. Automating entire clinical and administrative pathways transforms capacity.

Agentic automation in practice: freeing clinical capacity

An AI imaging vetting agent demonstrates the capability of agentic automation. This system can operate as a true end-to-end agent, not a collection of isolated tools. It intercepts imaging requests from both internal trust systems and GPs, understands the clinical content regardless of format or structure, identifies duplicate requests, extracts relevant data and clinical indicators, and compiles a comprehensive assessment of imaging appropriateness. The agent then validates this assessment against both national guidelines and local trust policies.

Crucially, the agent makes autonomous decisions within its defined scope. Appropriate requests proceed directly to scheduling – the agent books the investigation appointment without human intervention. Inappropriate requests are automatically rejected, and detailed explanations are sent to the requester. Borderline cases flagged as ‘requiring radiologist review’ are forwarded to clinicians via e-forms, complete with all relevant context compiled.

Thousands of clinical hours could be redirected to clinical care and diagnostic reporting.

This is agentic automation in its fullest expression: autonomous operation, intelligent decision-making, human-in- the-loop when needed, and governance throughout.

Orchestrating complexity: clinical coding

Clinical coding presents another area where agentic capabilities could prove essential. NHS trusts are funded based on accurate coding, making this critical to their financial sustainability; yet, recruitment in this specialised area remains a persistent challenge. The complexity lies not in any single task but in the orchestration required: extracting clinical information from multiple sources, applying coding rules, auditing for accuracy, and seeking clarification when needed.

AI agents extract relevant clinical information from notes and records using natural language processing. They apply coding rules and generate diagnostic and procedure codes. But rather than simply outputting results, the system orchestrates an audit workflow. Results for more complex spells are reviewed, quality-checked by the coders, and – when there is uncertainty or missing information – referred back to the originating clinician for clarification before the coding is finalised. Only then does the code enter the trust’s encoder systems.

This orchestrated approach drives significant productivity whilst maintaining the accuracy and compliance that clinical coding demands.

Navigating the integration challenge

One of the NHS’s most persistent obstacles is technical fragmentation. Multiple legacy systems, including patient administration systems, electronic patient records, referral platforms, maternity systems, investigations systems, and scheduling systems, exist across trusts and integrated care systems, often struggling to communicate effectively.

Agentic automation bridges these gaps without costly system overhauls. By acting as an intelligent integration layer, automation platforms enable data flow between disparate systems, extracting information from one format and presenting it appropriately to another.

The governance question also looms large in risk-averse healthcare settings. Clinical safety, data privacy, cyber security and public accountability are non-negotiable. Effective automation platforms must provide comprehensive audit trails, monitor AI model behaviour to reduce hallucinations and drift, and maintain compliance with evolving AI regulations and clinical safety.

Building sustainable, people-centred change

The success of automation depends fundamentally on people – empowering staff, building digital confidence, and embedding continuous improvement cultures. Productivity gains prove most sustainable when combined with workforce transformation that recognises automation as augmenting rather than replacing human expertise.

Scaling remains perhaps the greatest challenge. Many NHS organisations pilot automation successfully but struggle to embed it sustainably. Key enablers include executive sponsorship, dedicated automation centres of excellence, and willingness to share successful implementations across trusts.

The next frontier: orchestrated care pathways

The future of agentic automation lies in orchestration across entire care pathways. Rather than merely processing referrals faster, future agents could orchestrate complex patient journeys by identifying patients likely to need intervention before acute episodes occur, coordinating preventive care across multiple providers, managing follow-ups and monitoring, and escalating to clinicians only when their expertise is truly essential.

It means moving beyond isolated automation projects to connected ecosystems where AI agents, human expertise and digital systems work in concert across organisational boundaries.

The question facing NHS leaders is not whether to automate, but how quickly they can implement proven solutions that free clinical staff to do what they do best: care for patients. When technology, people and process improvement align, the NHS can transform productivity challenges into opportunities for sustainable improvement, delivering better access, improved outcomes and more efficient resource use at precisely the moment when all three are needed most.

References

1. BMA. (2025). NHS backlog data analysis. Retrieved from https://www.bma.org.uk/advice-and-support/nhs-delivery-and-workforce/pressures/nhs-backlog-data-analysis
2. NHS England. (2025). Waiting list falls as NHS staff treated record numbers last year. Retrieved from https://www.england.nhs.uk/2025/02/waiting-list-falls-as-nhs-staff-treated-record-numbers-last-year/
3. House of Commons Library. (2024). Backlogs in the NHS. Retrieved from https://commonslibrary.parliament. uk/research-briefings/cdp-2024-0181/

Please Note: This is a Commercial Profile

This work is licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International.