About This Case Study
This is an internal case study. Unlike most stories on this page, the client here is Grassroots IT. We believe that to genuinely help our clients get more from their technology, we need to walk the same path — identifying inefficiencies in our own business, then solving them with the same tools and approach we bring to client engagements.
This story is about what happened when our Service Delivery and Digital Solutions teams turned their attention to one of our own recurring frustrations: recruitment. It’s a practical example of what’s possible when you start with a business problem and let the right technology follow.
The Challenge
Recruiting for a specialist IT role is hard enough. But for James Keinzley, Service Delivery Manager at Grassroots IT, every recruitment drive came with a hidden problem — one that had been quietly costing the business for years.
Grassroots IT had been working with a trusted external recruiter to source candidates — someone who knew the business well and had a strong feel for the culture. But high-volume recruitment is a challenge for anyone. In one recent round, more than 600 people applied for a single role. Of those, roughly 300 were never screened at all — not a reflection of anyone’s effort, but simply a reality of the sheer volume involved.
“For us, what it meant was 50% of the possible candidates didn’t even get a look in,” James explains. “And considering we were struggling to find people that were good for the role at that particular time, we were really concerned that we’d just miss really good candidates purely because they ended up on the wrong side of the draw.”
This wasn’t a new problem — it’s one that affects businesses across the board whenever candidate volumes spike. The process was sound: brief the recruiter, trust her filter, receive a shortlist. But with no visibility over the broader candidate pool, the team had no way of knowing what they were missing.
With a new recruitment drive approaching — and a particularly specific brief — the team decided it was time to do something about it.
The Approach
The idea came out of a Level 10 leadership meeting, where the team was reviewing their recruitment strategy. They considered several alternatives — running their own LinkedIn campaign, creating a dedicated careers page, or bringing the recruitment entirely in-house — but ultimately decided the funnel itself wasn’t the issue. The screening was.
“We decided it’d be better to still keep the same funnel, but apply a different filter to it,” says James. “That’s when we had the idea of running something through AI.”
Initially, the team explored a Microsoft Copilot-based solution. James had already started manually copying and pasting resumes into Copilot to assess them — a process that worked, but was time-consuming and still required significant manual effort. Sulaiman Arshad (Sully), a Digital Solutions specialist at Grassroots IT, saw a faster path.
“It was an issue that kept coming up and being discussed repeatedly,” Sully recalls. “And one day I said, I feel like we can set this up pretty quickly in Power Automate.”
“Any manual steps that people are taking, they should stop and think — can this be automated? Because that can significantly improve any business’s efficiency.” — Sulaiman Arshad, Digital Solutions Specialist
The key insight was to treat it as a business problem first and a technology problem second. Rather than fitting the workflow around a specific tool, Sully assessed what the process actually needed — consistent, criteria-based screening at scale — and selected Power Automate as the right platform to deliver it.
How It Works
The solution is built entirely within Microsoft Power Automate, using tools that Grassroots IT’s clients already have access to as part of their Microsoft 365 licences.
A dedicated careers email inbox was created as the single destination for all applications. From there, an automated flow handles everything.
When a new application arrives, the flow:
- Detects any attachments (resume and/or cover letter, in PDF or Word format)
- Extracts the text
- Runs an AI prompt against the extracted text, assessing the candidate against the job description and a set of defined criteria
- Categorises the application as Acceptable, Maybe, or Hard Pass — with a tagged reason for each decision
- Moves the email automatically into the corresponding folder in the inbox
The AI component is powered by Microsoft AI Builder — a set of pre-built AI models that sit natively inside Power Automate and draw on the same underlying technology as Microsoft Copilot. Rather than connecting to an external AI tool or requiring any custom model training, Sully was able to configure an AI prompt directly within the flow. The prompt reads the full text of each application — cover letter and resume combined — and makes a reasoned judgement based on the criteria it has been given. It doesn’t just keyword-match; it interprets context, weighs up alignment, and produces a category and a reason, the same way a person would — just in seconds, and consistently, across every single application.
The AI prompt was built using the job description as its foundation, with additional hard-pass criteria layered on top — for example, candidates must be based in Australia or New Zealand and hold the right to work. Candidates in the Maybe category are reviewed manually before a final decision is made.
The tagging feature — which labels each application with a specific reason such as “strong MSP match” or “tech stack mismatch” — was added by Sully after observing that team members were applying their own notes manually. It was a small but significant improvement that made reviewing faster and more consistent.
The entire build was able to be completed by Sully within one day.
The Results
The impact was immediate. Of the first five interviews James conducted using the tool’s shortlist, every single candidate was someone he would have hired.
The tool has shifted where time is spent, and that shift matters. Around eight hours of manual effort were recovered in the first week alone — time that had been spent copying and pasting applications into Copilot by hand. More significantly, every hour James now invests in the process is focused entirely on strong candidates. There’s no wading through unsuitable applications, no uncertainty about what might have been missed. The screening work is done before he even opens an email.
“Every recruitment run I’ve done so far has been tedious and time consuming. But this time, the time I’ve been spending has actually been more meaningful.”
There was also a result nobody had anticipated. One candidate was flagged as a Maybe — not because of a skills gap, but because the AI identified them as significantly overqualified. The candidate had held high level management roles at global corporations, with salary expectations well beyond the role on offer. The system had spotted a risk that James hadn’t even thought to build into the criteria.
“That was probably the most surprising thing,” James says. “That it actually picked up on overqualified, as opposed to just underqualified. That was something I hadn’t even thought about, let alone built into my logic.”
“Make the time to invest in the solution. You’ll never get time to put the energy into that if you don’t create the time. And the solution is what creates the time.” — James Keinzley, Service Delivery Manager
What's Next
The current tool was built as a proof of concept for a single hiring round. But the team is already thinking bigger. James and Grassroots IT’s leadership have discussed building a permanent “bench” — a careers page that stays open year-round, continuously collecting and screening applications across multiple roles, so that when a position does open up, qualified candidates are already waiting.
“We could use this to start building that database up,” James says. “Level ones, level twos, level threes, finance and admin — have this as a back-end recruitment machine running off our website.”
Sully sees similar potential for Grassroots IT’s clients. The base solution is already built. With some configuration around specific roles, screening criteria, and existing HR workflows, it could be adapted for any business managing high-volume recruitment.
Successful Outcomes
By building a purpose-fit automation using tools already available in Microsoft 365, the Grassroots IT team transformed a longstanding recruitment bottleneck into a streamlined, consistent process. Key outcomes included:
- 100% of incoming applications screened — no candidate left unreviewed due to volume
- AI-powered triage delivering consistent, criteria-based decisions across every application
- 5 from 5 interviews resulting in hireable candidates — a marked improvement on previous rounds
- Approximately 8 hours of manual effort recovered in the first week alone
Full solution built and live in under a day, using existing Microsoft 365 licences
The Broader Lesson
For James, the most important takeaway isn’t about the technology — it’s about the mindset required to get started.
“There are a lot of problems that businesses have where they think ‘there’s got to be a better way’. They don’t even know where to start — they start Googling, or they ask in Copilot. But it’s worth having a conversation with your MSP or your digital solutions team, and saying ‘I have a business problem, how can we solve it?’. Even a 15-minute conversation can lead to a solution like this.”
“We had no idea what the outcome would be or what the actual end technology solution would be. We just knew that we had a business problem. And by just having that discussion, we turned out a solution that didn’t take as much time as we thought it would to uplift the quality of what we’re doing.”
The lesson? The solution is almost always simpler than the problem feels. You just have to make the time to look for it.
Got a business problem that needs a better solution?
Talk to the Grassroots IT Digital Solutions team. You don’t need to know what the technology looks like — you just need to know what’s slowing you down. We’ll take it from there.
Learn more about our Digital Solutions services
