Case Study: LAP South West – Driving Efficiency Through AI and Automation

The Challenge

How can a multi-academy trust leverage automation and AI to reduce workload, streamline processes and improve educational outcomes without adding complexity?

The trust needed solutions that were: 

  Efficient and scalable across multiple schools. 
  Easy to adopt with minimal training. 
  Cost-effective given budget constraints. 
  Flexible to support both administrative and teaching needs. 

The goal was to move away from fragmented, time-consuming processes and create a coordinated approach that would support staff needs, improve resource allocation and facilitate personalised learning. 

This specific project began with a common issue. People and Payroll teams were receiving large numbers of repeat HR and payroll queries. Although the queries were valid, the volume slowed response times and created frustration.

Here’s LAP South West’s  Trust People Advisor, Emma Richards, explaining the challenges they faced:

“As support colleague pay awards have risen, staffing levels have reduced, and more administrative tasks have moved centrally or into automated systems. Colleagues were turning to leaders more often, increasing day‑to‑day questions and pressure on Headteachers.

“As our Trust grew, so did the number of queries. We wanted colleagues to access clear answers quickly, not in several days, but when they needed them.

“Accessibility was also important. We support colleagues across different schools, roles and levels of digital confidence. Any solution had to be simple, inclusive and easy to use. Accuracy and consistency mattered too. Colleagues needed trusted information every time. We also have a duty of care. Many HR queries involve wellbeing or personal circumstances, so the solution needed safe boundaries and clear signposting to human support.”

“And it had to be sustainable, scalable, and low-cost, reducing pressure on leaders and business support while allowing the People team to focus on conversations that need a person. These combined challenges led us to work with Computeam to explore whether a Microsoft agent could shift us from reactive email support to something more proactive, accessible and empowering for colleagues.” 

The Solution

The reality of UK education is that schools and trusts are both time-poor and budget-poor. Any solution worth recommending has to work within that reality, not against it.

For this agent to make a real difference, it had to be equitable, available to every member of staff rather than a select few. But reaching all 430 colleagues across the trust through full licensing would have come to around £5,800 a month. For most trusts, a figure like that ends the conversation before it starts.

As a client partner, our advice was to begin differently. We suggested a pay-as-you-go (PAYG) model as a way to get started, which is ideal when you want to trial an agent across the entire staff without committing to high upfront costs. The first month came to just 92p, plus the cost of a single licence for the staff member setting the agent up in Copilot Studio. To date, the agent has cost the trust £32.87 per member of staff.

Over time, the savings the agent generates will outweigh the running cost of the PAYG model, and those savings can be reinvested into staff licensing as part of a planned roll-out. When you're getting started, PAYG is the key to proving the value at low cost before moving to larger licensing commitments.

For Emma Richards, removing the cost barrier was what made the agent viable. But it was how closely it matched the trust's needs that convinced her. She adds:

“When the idea of an agent was explored, it became clear how well it matched what we needed, particularly in helping colleagues self-serve. We wanted a simple way for colleagues to access answers, guidance and key documents whenever they needed them. The agent gives us exactly that…”

The key implementation choices we made

First, and most importantly, we connected the agent directly to the LAPSW SharePoint policies and folders deliberately chosen by the Trust – getting this right was crucial. 

We didn't point the agent at everything, only at content that was agreed to be accurate, current and appropriate for colleagues and leaders to rely on. During the build, this forced clarity about exactly what the agent should be doing and responding to. In addition, if a document was out of date or unclear, the agent quickly exposed that, which helped improve the underlying content.

We also decided to disable web search entirely. This ensures that all answers are generated exclusively from internal LAPSW sources, giving colleagues complete confidence in the accuracy and security of the information. From a build perspective, that was crucial because it kept the agent focused and predictable, and avoided the risk of generic or conflicting advice appearing alongside Trust policy.

As part of the creation, we engaged the People & Payroll professionals, who are experts in their fields. They collated the relevant documents to embed in the agent and tested it extensively. These are areas where accuracy is critical, and where colleagues often feel least confident. Testing with real scenarios helped us refine the language, spot gaps and ensure the answers were usable in practice.

Before any wider rollout, we piloted the agent across a handful of school teams. The teams represented different roles and levels of confidence with policy and systems. Piloting in real school environments helped us understand how colleagues naturally ask questions. The feedback shaped the language, tone and structure of the answers, making the agent far more practical and usable.

Overall, the implementation choices focused on building confidence, trust, and usability from day one, so the agent feels like a natural extension of how LAPSW already supports its people.

Guiding colleagues with the right prompts

Building on the pilot work, we wanted to focus on how colleagues experience the agent when using it, and the role that alerts, messages and prompts play in that experience.

One of the key things we learned during the pilot across a handful of school teams was that the way information is presented is just as important as the information itself. Colleagues were approaching the agent with varying levels of confidence, urgency and understanding of policy, so we needed the agent to respond in a way that felt supportive, clear and reassuring.

Emma deliberately designed the suggested prompts to help colleagues get started quickly and confidently. The prompts reflect the most common People and Payroll questions, using simple, everyday language that mirrors how colleagues typically ask for help. 

From an implementation point of view, that reduced hesitation, guided colleagues towards trusted topics and reinforced the agent’s boundaries without being explicit. 

During testing, we saw that suggested prompts increased engagement and helped colleagues immediately understand what the agent could support. It also helped us think about the sorts of questions that might be asked and test those scenarios accordingly

Guardrails and safeguarding built in

For Emma, usability always had to sit alongside safety. She explains:

“Another key part of our build was putting strong guardrails in place using Microsoft Studio. Through the topics and instructions features, we were able to make sure the agent only gives trusted, policy‑aligned information, and never offers opinions or judgment." 

"The example below shows how the agent responds when someone mentions being bullied, signposting to support rather than trying to advise. Testing showed us exactly where the agent should stop and where a person needs to step in, which was essential given our duty of care. We have a duty of care to all our colleagues, and we wanted to ensure that safeguarding remained personal.

"To support our safeguarding systems, we built in alerts using the topics feature and Microsoft Flows. When sensitive words or phrases are used, such as concerns about health, well-being, bullying, or depression, the agent provides urgent support information and automatically emails the People team mailbox. This means we can pick up sensitive or urgent matters quickly and ensure colleagues receive human support when needed.”

For Emma Richards, this was where the technology and the trust's duty of care came together most clearly: 

“We were intentional about the messages and prompts colleagues see when they use the agent. It clearly explains what it can support, where its answers come from and when a human conversation is more appropriate. That clarity helped build trust during the pilot.”

“When a safeguarding alert is triggered, the People Team receives a notification that includes who raised the concern so we can step in quickly and offer the right support. This meant the agent was not only giving information, but also acting as an early safeguarding system, reinforcing that wellbeing and safeguarding remain human‑led.”

“These alerts, messages and prompts were a core part of implementation. By testing them in real settings and listening to feedback, we shaped an experience that feels safe, practical and genuinely supportive for colleagues across our Trust.”

Client Statement

“Colleagues are getting answers more quickly and independently, which builds confidence and reduces reliance on others. Speed and consistency have been two of the biggest immediate gains.”

“Leaders are also seeing fewer day‑to‑day queries. This clears the path so they can focus on what matters most, the teaching and learning of our children.

"Fewer repeat emails have helped ease inbox pressure, giving teams more time for complex or sensitive conversations that require human support.”

“Accessibility has been another clear benefit. The agent is available at any time, works across devices, and supports colleagues who find it easier to access information this way than by searching policies or asking directly.”

“And importantly, all of this has been achieved at a very low pilot cost, giving us strong proof of concept and a clear case for scaling in a sustainable way.”

“After sharing the results of our pilot, our Education team, Compliance and Health and Safety teams are now considering their own agents, and because the frameworks are already in place, additional agents can be configured and deployed quickly.”

“The agents also give us valuable insight into search patterns and common queries. This helps us strengthen our systems, improve communication and identify where processes or documents need refining.”

“We can now share our model with other Trusts and partners to adapt in their own contexts. Faster access to clear guidance frees colleagues to focus on supporting children and families. “

“We may work in different educational establishments, but we all hold the same responsibility to provide safe, consistent and high-quality experiences for every child. By sharing our approach, we are strengthening colleague support across the system and improving conditions for children to learn and thrive.”

Emma Richards, Trust People Advisor, LAP South West

Computeam Statement

"Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. "

Computeam