Rethinking unnecessary complexity with AI

“Genius lies in simplicity”
– Mark Twain

For Liquid Partner Steven Marrinan, one of the architects behind Liquid’s AI Leap Workshop, simplicity is not the starting point.

It is the “light bulb” moment organisations walk away with, post-workshop, realising intractable problems and pain points can be solved with the support of fit-for-purpose AI systems.

Not general-purpose tools that have flooded the market since the advent of ChatGPT, but enterprise systems, intentionally co-designed for safety, efficiency and reliability, complete with human-at-the-centre control layers.

Understanding the right problems

As Steven explains, the true value of the AI Leap Workshop lies in collaborative problem-solving: working with organisations to understand operational pain points; identify where complexity is hindering efficiency; and determine how AI can be safely and responsibly scaled across workflows.

“Organisations come to us because they know there’s something they should know, but they can’t really pinpoint it, and that’s understandable when so much of the AI conversation feels like it swings between two extremes of hype and hesitation,” Steven says.

“The workshop provides clarity and a solid entry point for leaders and teams to understand where AI can make a genuine and measurable difference, grounded in practical day-to-day use cases.”

Ready to take the leap? Start a conversation today

Asking the right questions

Launched in the wake of Australia’s first national survey into public sector attitudes towards AI, conducted by Liquid in collaboration with The Mandarin, the AI Leap Workshop primarily targets government agencies and, more widely, assurance-focused organisations (for example, in healthcare and education).

It is structured as a sequence of short but intensive stages, pre-workshop planning (questionnaire, draft map of an existing process or pain point, related data sources) and in-person session, with a focus on identifying real opportunities for AI systems to:

reduce administrative burden;

improve service delivery;

strengthen compliance; and

support (not replace) expert human decision-making.

Get a head start: download Liquid’s practical AI project proposal and governance checklists [LINK to checklists]

Easing #1 public sector pain point

Not surprisingly, a key focus of AI Leap Workshops to date targets administrative burden: identified as the biggest pain point across the public sector in Liquid’s Let’s get real about AI survey. Read full report (released August 2025).

When asked where AI investment should be prioritised to achieve the greatest impact in the public service, 42% of 600 surveyed participants answered “reducing administrative burden for staff”.

Further, in their Top 5 operational challenges, 409 participants ranked “reducing the manual workload on staff so they can focus more on complex or sensitive cases” at No. 1.

Coming in a close second, 401 pointed to the bugbear of siloed information and inaccessible data, stating “information is spread across multiple systems that do not talk to each other”; while, at No. 4, 364 said “staff spend too much time on repetitive tasks like data entry or document handling”.

Meeting organisations where they are

For Steven, the findings provide clear direction on the need to meet organisations where they are, recognising the urgent call for practical solutions to ease administrative burden so people can focus less on process and more on outcomes. Read his recent article in The Mandarin, How AI can lift the weight of public sector administration.

As he sees it, administrative burden is the area where AI can deliver the most immediate benefits. Fast.

Left unchecked, administrative burden can manifest in a range of ways: creating operational inefficiencies that impact service delivery and compliance outcomes, while also taking a toll on workplace wellbeing, ultimately leaving organisations vulnerable to avoidable risk.

“Administrative burden is enormous and plays out with all the paperwork and document trail that has to be recorded in government because everything has to be accountable to something else,” Steven says, referring to “a pattern of assurance” that characterises the public sector, anchored to rules and legislative requirements.

Based on recent workshop outcomes, he adds that administrative pain points are often rooted in structural problems, rather than a lack of expertise, effort or access to the right information (aka structured and unstructured data that forms any evidence base) and institutional knowledge.

Instead, the real problem turns on extreme manual effort required to locate, assemble and utilise information prior to a decision.

This is precisely where AI-enabled process redesign can have the greatest impact. Value is created not by replacing people but redesigning how work is organised and executed to support human decision-making.

The right AI tools for the right job

“Anything that currently involves repeated manual effort to create assurance, especially where the criteria and evidence remain the same across a decision cycle, and confidence is achieved from human repetition, is a good candidate,” Steven says.

“The great part of the workshop is being able to say, what if we can take all this administrative burden away from you and build a trusted system that you’re still very much in control of, so you get the best of both worlds?

“And we do go through sort of emotional journeys with organisations to arrive at that light bulb moment, when people realise there’s a better, simpler way.”

Steven points to an example, unearthed in an AI Leap Workshop, of a high-friction validation process in the regulatory space that could be immediately and responsibly solved by applying the right AI tools to absorb repetition at scale, freeing skilled people to focus on oversight, higher-value decision-making and quality.

“In this instance, the process involved up to seven checks of the same evidence with the same criteria, creating assurance through repetition of manual work,” he says of a process related to relatively low-judgement but high-assurance decisions about payment of individual invoices linked to high-volume, fixed work.

“In future, if the same process needed more assurance, they would probably add more people and manual checks, which is not sustainable or productive.”

Of note, this checking and re-checking rarely altered the outcome, but increasingly sapped the energy of highly skilled staff.

To address this, the AI Leap Workshop provided a tailored blueprint for responsible application of an enterprise system (not one-size-fits-all platform), enabling evidence to be assembled once, decision logic to be applied consistently and controls to be embedded directly into workflows.

Taking away repetitive tasks

“This rules-based pattern of assurance, which has historically been managed by more and more people and more and more checks, is now completely solvable,” Steven says, noting parallels between invoicing, grant applications, auditable safety inspections and more.

“Similarly, in healthcare, do we want to have nurses spending as much time on administration or is it better for them to be able to spend more time doing the job they’re trained for… caring for people?

“AI is just another technological tipping point. If you point it at the right problem, with the right design and controls, it will do the job it’s suited to, replacing repetitive tasks that most people hate. It’s about augmenting not replacing human capability.”

Introducing Unearth AI

AI works best when embedded into workflows in a way that complements existing roles and responsibilities.

Liquid’s sovereign AI platform, Unearth AI, makes this possible.

As an Australian-made, locally hosted solution, Unearth provides a foundational model that moves beyond isolated solutions to deliver consistent, scalable capability. Find out more

Take the leap and start a conversation today

Rethinking complexity with AI

Sign up to our newsletter.