Articles

The AI-powered future for rostering vendors

In the NHS, each local trust has its own way of working, especially when it comes to rostering staff. However, within each trust, there may also be variations from one unit to the next, which can often create big headaches for rostering software providers.

I remember examining one trust’s rules on preventing staff from working too many long days in a row. These rules keep patterns safe, but they aren’t one-size-fits-all and often require a more nuanced approach. For example, a high-intensity inpatient ward needs much tighter guardrails than, say, a 9-5 community service.

And that means everything can get very complex, very quickly.

Until now, trusts have frequently needed to find workarounds to accommodate such scenarios, since most software providers don’t offer this level of configurability – largely because older legacy systems simply can’t manage that kind of complexity.

But, recently, we’ve been asking ourselves: do we need to be so rigid? Particularly because the integration of AI into software development processes is opening up a whole new world of possibilities that can help providers solve challenges like these.

So, we’re now experimenting with automation that uses AI to let units set their own workflows without needing a developer to change the core code. The goal here is to build a system flexible enough to handle such demands – turning what used to be a technical blocker into a genuine innovation.

Leaning into the challenges

The kind of challenge and solution outlined above is just one example of the way AI is starting to become a powerful problem-solving tool for rostering software providers.

In our experience, the most common challenges are:

  • Inherited technical debt: Providers often find themselves dealing with outdated legacy systems or needing to re-platform the whole system. This can feel like that well-used analogy of changing the engine on an aircraft while it’s still flying.
  • Integration and interoperability: To reduce complexity and ease end users’ workloads, it’s vital for rostering systems to integrate seamlessly into other operations, such as HR, finance/payroll and compliance.
  • Security and compliance: Since many end users are working in essential services, getting this right, so it’s easy to ensure staffing levels are always correct, is non-negotiable, both ethically and legally.
  • End-user reassurance: Typically, administrators can find it hard to visualise the solutions they need, or may have little trust in technical solutions as they’ve been let down previously. It’s vital, therefore, to have the capability to work closely with users to give them the confidence they need.
  • Scaling quickly and effectively: Often rostering platforms will need to scale rapidly – which can be highly challenging if they’re built on outdated or niche technology.

These real-world headaches mean that, all too all too frequently, many end users have to struggle with inflexible, underperforming tools or the burden of manual admin, while managing unpredictable demand, unforeseen scenarios and frequent last-minute changes.

A better way of rostering

Based on our experience of working closely with rostering software providers, we have a clear picture of these major pain points and how such issues are all too common across many industries.

But we also know there’s another way, where AI tools can provide simple, rapid and adaptable solutions that allow workforces to focus on their core competencies rather than hours of admin.

In general, effective, future-ready rostering must therefore be highly configurable, in order to adapt to those varying requirements. It must be able to flex to sector-specific constraints. And it must support different models – such as filling time-based slots (perhaps hospital shifts) versus filling service needs (such as when planning a higher education institute’s timetable).

To achieve outcomes like this, AI tools offer the opportunity to massively shorten feedback loops when developing and iterating new versions of existing systems. It’s even possible for providers to undertake rapid prototyping and validation in a live environment. This could involve working directly with end users to build multiple iterations of a system in order to demonstrate exactly how a solution will work in practice.

Additionally, AI’s ability to analyse vast data sets can deliver valuable predictive analytics, such as an understanding of when staff absences or customer/patient demand are most likely to be at their peak. Based on historical data patterns, it can make intelligent suggestions for administrators to choose from.

Or, as in the example at the beginning, AI’s capacity to take on previously impossible challenges enables developers to solve major technical problems that previously would have been considered far too time-consuming and financially unfeasible. An AI-powered tool can also help to ensure rostering is done as fairly as possible, without any human biases. With a good dose of transparency, it can show staff how and why rostering decisions are made so no one feels they are being singled out for unfair treatment – for example, by demonstrating that everyone will end up working an equal number of public holidays throughout the year.

Human accountability

Despite the undeniable advantages that AI can bring to rostering tools, for the foreseeable future there will always need to be human involvement throughout the process. The overriding aim should be to improve productivity by reducing development or admin time, rather than to replace human judgement.

In the first instance, we would always recommend that AI features be introduced gradually into rostering tools and with full transparency – not only to ensure maximum effectiveness, but also to help build trust among users.

Before moving to anything near full automation, a tool should offer explainable alternatives to users – for example, by suggesting several options with a rationale of why it has proposed each one. It would then be up to the human administrator to select the most appropriate, all the while helping the AI to learn what works best in that specific organisation’s context.

Transparency and explainability are also crucial when it comes to ensuring all outputs lead to full compliance, particularly in sectors like healthcare and education, where regulatory observance and safety are critical.

The path forwards

The future of AI-powered, demand-driven rostering and rostering software development – using systems that are accurate, compliant and sensitive to end users’ needs – is now within reach.

To get there, it requires a highly collaborative approach between end users and software developers who have in-depth product and domain knowledge and are powered by expert engineers with advanced AI skills.

To find out more about how we can help you remove the pain from rostering software development with our AI-powered solutions, get in touch now.

Iskra Ristovska, Principal Product Owner at Damilah