How a Brisbane Recruitment Agency Placed 4 More Candidates Per Month

It's 4:15pm on a Thursday and Tom Nguyen is staring at a spreadsheet that's been open since Monday. Not the same tab — he's been updating it all week — but the same fundamental problem: 4,200 candidates in Apex Recruitment's JobAdder database, and somewhere in there are the right people for three open construction roles his clients have been waiting on for nearly a month.

The roles aren't obscure. Senior site supervisor, project engineer, and a civil foreperson for a major Brisbane infrastructure project. Good salaries, excellent clients, genuine opportunities. The candidates almost certainly exist in the database. But finding them, re-engaging them, checking their availability, scheduling interviews, collecting references — that's the part that takes forever.

Tom runs Apex Recruitment from the CBD, four consultants specialising in construction and engineering placements across South East Queensland. It's a tight, high-quality team. But tight teams don't scale without automation, and by early 2026 the admin load had quietly become the agency's ceiling.

"We weren't dropping balls — the work was getting done," Tom told us. "But every consultant was spending three hours a day on things that weren't actually recruitment. Screening, scheduling, chasing references. We were good at the job; we just couldn't do enough of it."

The Four Drains on Consultant Time

When we mapped Apex's workflows before the OpenClaw deployment, four bottlenecks stood out as accounting for the majority of wasted consultant hours.

1. CV Screening at Scale

A typical role brief at Apex generates 80–120 applications via Seek and LinkedIn. With four consultants each running multiple roles simultaneously, that translates to hundreds of CVs being processed every week — mostly manually. A consultant would open each CV, scan for the relevant credentials and experience, shortlist or reject, and log the outcome in JobAdder. For a 100-application brief, that's two to four hours of work before any actual recruitment has happened.

"We'd tried some of the ATS built-in screening tools," Tom says. "They're blunt instruments. They keyword-match and miss half the relevant candidates while surfacing half the irrelevant ones. We ended up not trusting them and doing it manually anyway."

2. Interview Scheduling: Six Emails to Confirm One Meeting

Getting a candidate and a hiring manager into the same room — or on the same call — sounds simple. In practice, it's one of the most tedious coordination tasks in recruitment. Apex consultants were averaging six emails back and forth to confirm a single interview. Multiply that across twenty active roles and four consultants, and you're looking at hours of scheduling per week that adds zero value to the placement itself.

3. Reference Checks Taking Up to Five Business Days

Reference checks are a regulatory requirement — the RCSA Code of Professional Conduct is explicit about the obligation to verify candidate credentials, and under the Fair Work Act, employers have legitimate interests in knowing who they're hiring. But the operational reality of collecting references is relentless. Referees don't answer calls. Emails get lost. Follow-ups get forgotten. A process that should take a day was routinely taking three to five business days at Apex, adding weeks to the overall time-to-fill on competitive roles.

4. A Database That Wasn't Working

The 4,200 candidates in Apex's JobAdder database represent years of relationship-building, interviews, and placements. But 95% of them had never been re-engaged after their initial placement or application. These were qualified, verified people who'd already been through Apex's process — and they were sitting idle. Meanwhile, consultants were advertising on Seek for candidates who might already be in the database and available.

Under the Privacy Act 1988, candidate data can only be used for the purpose it was collected — broadly, for employment placement. But re-engaging previously placed or assessed candidates for new relevant roles is explicitly within scope. The data was legally usable. It just wasn't being used.

Finding OpenClaw

Tom wasn't new to the idea of automation. He'd tried three different recruitment-specific software tools over the past two years and found the same pattern each time: promising demos, painful onboarding, and a product that solved one specific problem while ignoring everything else.

"Every tool was built for a specific task," he says. "One was great for job advertising. Another had a decent scheduling widget. None of them talked to each other, and none of them understood the full recruitment workflow."

A contact at a RCSA Queensland chapter event mentioned King Klaw in early 2026. What caught Tom's attention wasn't the AI pitch — it was the integration. King Klaw deploys OpenClaw, an open-source AI platform, and the local team had built a native JobAdder integration that sat across his entire workflow rather than replacing any one part of it.

"The framing was different," Tom says. "It wasn't 'here's a tool that does X.' It was 'tell us how your business works, and we'll make AI fit into it.'"

He booked a scoping call. By the end of the first session, they'd mapped every manual touchpoint in Apex's candidate lifecycle. By the end of the week, they'd scoped a deployment.

The Deployment: Integrated Across the Full Workflow

The King Klaw team spent the first week mapping Apex's existing processes in detail before writing a single line of configuration. The goal wasn't to automate for automation's sake — it was to remove the mechanical work around the decisions that only Tom's consultants could make.

The Privacy Act compliance layer was built into the re-engagement sequences from the start. Every outreach included a clear statement of purpose (employment placement), an opt-out link, and automatic suppression for any candidate who had requested their data be removed. The system maintains a compliant audit trail of all candidate communications — relevant for any RCSA code obligations around candidate management records.

Tom also notes that the Fair Work Act's general protections provisions were front of mind during setup. The AI screening tool was configured with explicit criteria from each role brief — not generalised demographic or personal attributes — so that shortlisting decisions were defensible and consistent. Consultants review every AI-generated shortlist before any candidate is contacted.

The Results: One Month In

We spoke with Tom at the 30-day mark. The numbers were clearer than either of us had expected.

CV Screening

The first role brief run through the new system was a 100-CV batch for a senior project engineer role. The AI shortlisted 8 candidates with a written rationale for each — experience match, qualification verification, availability flag, and any gaps for the consultant to probe at interview. Total processing time: 20 minutes. The same task had previously taken one consultant a full half-day.

"I went through the eight with our consultant and agreed with six of them immediately," Tom says. "The other two we decided weren't quite right once we looked at them properly — but they were reasonable inclusions. That's a much better hit rate than we were getting doing it manually."

Interview Scheduling

Average time from shortlist to confirmed interview dropped from three days to four hours. The self-scheduling link eliminates the back-and-forth entirely — candidates see the consultant's real-time availability and book directly. Confirmation and reminder messages go out automatically. Cancellations trigger an immediate re-booking prompt.

Reference Checks

The structured email-and-SMS reference sequence achieved an 80% completion rate within 24 hours in month one. The remaining 20% were followed up automatically at 48 hours. The previous average of three to five business days has effectively been eliminated as a time-to-fill variable.

Database Re-engagement

This was the result that surprised Tom most. In the first month, the AI matched open role briefs against the 4,200-candidate database and sent personalised outreach to 340 relevant candidates. Eight of those conversations converted to new placement opportunities — roles that would otherwise have required fresh advertising and a full new candidate pipeline.

"These are people we already knew," Tom says. "We'd placed some of them before. We just hadn't had the bandwidth to stay in touch properly. The AI did what a good consultant would do if they had unlimited time."

The Revenue Impact

Four additional placements per month sounds like a modest number. In construction and engineering recruitment, where average placement fees sit between $10,000 and $15,000 per role, it's $48,000 in additional monthly revenue. That's $576,000 annualised — from a team of four consultants who aren't working longer hours.

The time-to-fill improvement also has its own revenue logic. Cutting from 28 to 21 days means each consultant can turn roles faster — which means more roles can be run in parallel without the bottleneck of stalled applications. Tom estimates that the scheduling and reference-check automation alone freed up roughly 12 consultant-hours per week across the team. That's the equivalent of adding a part-time staff member without the overhead.

"I've been in recruitment for eleven years," Tom says. "I've seen a lot of tools promise efficiency gains and deliver marginal improvements at best. This is genuinely different. We're doing work at a higher velocity without any reduction in quality."

The OpenClaw Features That Drove the Results

The March 2026 OpenClaw release included several updates that were particularly relevant to Apex's deployment.

JobAdder Native Integration

OpenClaw's integration with JobAdder isn't a simple data sync — it's a two-way connection that allows the AI to read role briefs, access candidate profiles, update application statuses, log communications, and trigger workflow actions directly within the ATS. Consultants never leave their existing system. The AI operates inside it.

Structured Candidate Scoring

The CV screening model doesn't use a black-box relevance score. It generates an explicit rationale for each shortlist decision, tied to the criteria in the role brief. This is important for two reasons: it lets consultants verify and override the AI's judgement, and it creates a defensible record of how shortlisting decisions were made — relevant under the Fair Work Act's general protections provisions, where discriminatory screening is a live legal risk.

Multi-Channel Outreach Sequencing

The candidate re-engagement campaigns use OpenClaw's multi-channel sequencing skill — email, SMS, and LinkedIn message drafts generated in the consultant's voice and customised per candidate based on their profile history. The March 2026 update improved the personalisation logic significantly, producing outreach that reads as individual rather than mass-broadcast.

Compliance Audit Trail

Every candidate communication — outreach, opt-out, confirmation, reference check — is logged with timestamp and content in a structured audit trail. This satisfies the Privacy Act's requirements around data use transparency and makes RCSA compliance reviews straightforward. It's the kind of record-keeping that good agencies know they should have and rarely do consistently.

The Honest Take

Tom is candid about what the system doesn't do. The AI doesn't make placement decisions. It doesn't build client relationships. It doesn't negotiate offers or handle the human complexity of a candidate who gets cold feet the week before their start date.

"All the things that actually make a good recruiter — the relationship stuff, reading people, knowing when to push and when to back off — none of that has changed," he says. "What's changed is that my consultants spend their time doing that, instead of copying and pasting emails and chasing referees."

He also notes that the AI-generated CV shortlists required calibration in the first couple of weeks. The model needed to learn Apex's specific quality bar for construction and engineering roles — not just keyword matching, but understanding the difference between a site supervisor who's managed twenty-person residential builds and one who's run infrastructure projects. That calibration happened through consultant feedback, and the shortlist quality improved noticeably by week three.

"It's not a set-and-forget system," Tom says. "You have to engage with it. But the engagement is quick — reviewing a shortlist rationale takes ten minutes, not four hours — and the system gets better the more you use it."

What Other Recruitment Agencies Should Know

The pattern we've seen at Apex repeats across recruitment agencies of all sizes: the bottlenecks aren't in finding good candidates or building client relationships. They're in the mechanical work that surrounds those activities — the screening, the scheduling, the chasing, the logging.

Under the RCSA Code and the Privacy Act, recruitment agencies already have obligations to manage candidate data responsibly and maintain quality processes. Automating those processes — with the right compliance layer built in — doesn't create additional obligations. It makes meeting existing ones easier and more consistent.

The database re-engagement result is the one that should make most agency principals pause. If you have a database of qualified, verified candidates that you're not systematically re-engaging, you're leaving placement revenue on the table every month. The Privacy Act permits re-engagement for employment placement purposes. The technology now makes it practical to do at scale without a dedicated relationship management function.

Tom's advice: "Start with one workflow — just the CV screening, or just the scheduling — so you can see how it works with your specific roles. Once you've got that running well, the rest follows naturally."