AI In Practice: Airtasker, Kinatico & Pureprofile

Re-published to correct typo in story title.

On Monday, Research as a Service (RaaS), in cooperation with David Tasker and team at Chapter One Advisers, organised an AI in Action webinar, with the following CEOs from three ASX-listed small caps presenting:

• Tim Fung – Founder & CEO, Airtasker ((ART))
• Michael Ivanchenko – CEO, Kinatico ((KYP))
• Martin Filz – CEO, Pureprofile ((PPL))

Below is FNArena’s curated impression from all three presentations given under the monniker “AI in Action, How ASX Companies Are Leveraging Artificial Intelligence”.

Looking For Tangible Outcomes

Finola Burke, RaaS Group: Our thinking behind this webinar is that AI has brought a lot of uncertainty into the equity markets, and while we can’t predict where AI will take us, we thought it was worthwhile looking at how companies that RaaS covers are dealing with the threat, the challenges and the opportunities AI presents.

It’s our view smaller companies tend to be nimbler with technological change because management is closer to the coalface and can gauge and adjust their businesses accordingly.

It’s also our view some business models have moats which can help with the onslaught of AI.

David Tasker (Host): AI is clearly one of the dominant themes in markets right now, but the key question for investors is simple: where is AI driving outcomes, not just experimentation, but real impact on revenue, cost, product and ultimately earnings?

What makes today’s session valuable is that we’re looking across three very different business models; a marketplace, a regulated compliance platform, and a global data and insights business.

How does AI enhance a marketplace like this, rather than disrupt it? To walk us through that, I’ll hand over to Founder and CEO of Airtasker, Tim Fung.

Airtasker: Widening The Business Moat

Airtasker CEO, Tim Fung: Airtasker is building the world’s most trusted marketplace to buy and sell local services. It is a really simple business model; we connect people who need work done with people who want to work.

Our mission at Airtasker is to empower people to realise the full value of their skills. For us, creating jobs for humans isn’t a by-product of the work that we do, it’s the core purpose.

We’ve done over a billion dollars in jobs now globally and have now passed over 5 million jobs completed in the Australian market, and we are carving out a place in this marketplace ecosystem.

I think one of the things that’s interesting about AI is it’s going to make productivity gains really powerful in the white collar space.

It’s a little “un-intuitive”, I guess, to us from 10 to 15 years ago, but AI is making things like white collar jobs, and jobs that can be done virtually and remotely, really disruptive in that space.

Some 95% of Airtasker’s jobs require physical and real world skills, humans doing things with their hands, things like cleaning, moving, furniture assembly, and handyman services.

As Jensen Huang, the CEO of Nvidia, says: it’s really going to be the plumbers that win the AI race.

Anthropic has done some amazing research and released a Labor Market Impact Study. Jobs like management, business and finance, computer and maths, architecture and engineering, are the jobs that are going to be impacted here, with close to 100% of those jobs potentially being covered by AI.

Jobs like software engineering are already seeing impacts.

Real world jobs have much lower exposure to automation. Things like handyman work, insulation, and garden maintenance are much less likely to be automated.

There is also a frontier of humanoid robotics, which I think is worth calling out. We do think over time, humanoid robotics could become something that is impactful on labour markets.

That said, we always believe humans are going to be on the frontier of creativity and the things that are going to be done to further humanity.

Although some of these jobs, even in the physical world, could be disrupted eventually, new jobs are going to be created that we can’t even imagine right now.

I think we are some way away from humanoid robots taking all those jobs because they require a huge amount of physical, in person dexterity. That said, there’s also the disruption of Airtasker, or the potential disruption of various marketplace platforms and connectors.

One of the things I think is worthwhile really looking at is the difference between those that are going to come out with stronger moats post AI disruption, and those that are going to come out with weaker moats.

There’s a great article by Nicholas Bustamante that talks through ten moats that are very common in businesses.

The first list of potential moats in businesses include learned software interfaces, specific business logic that is created, or access to data which is ultimately publicly available. Those moats, because of the unlimited intelligence that you can get from AI agents, are going to be largely weakened moats.

There are moats in businesses which, in an agentic AI world, are going to get even stronger. Examples of that would be proprietary data, regulatory lock-in, and, of course, network effects; one of the things that Airtasker is built on.

Three of these are key moats.

The first is network effects. Airtasker is a marketplace. It is important when you come to Airtasker and post your job, you get the best quality offers, the best in the fastest possible time, with the biggest range of possible services.

We believe that’s driven largely by having a highly liquid base of customers and taskers.

The second is embedded transactions. Airtasker manages the payment.

If you look at some of the lead generation websites, which tend to broker information, that is, connect one person with another, I think those can be very largely disrupted.

Airtasker, in contrast, takes the payment, provides the insurance and the assurance and the system of record, which makes sure that those services are provided to the level expected.

The third thing is proprietary data. Airtasker is sitting on over 10 million reviews from customers, which are real insights into which tasker is going to be best for which kind of job.

By collecting that information and hosting that information for each of our taskers, we’ve created a unique piece of inventory which is very difficult to find anywhere else.

As agentic AI makes the economy more productive, you’re going to see more of a need for accessing that kind of information and that kind of inventory.

I think that’s going to bode very well for our business.

It’s also interesting to look at not just Airtasker as a business model.

We started by looking at the services economy. As a starting point, humanoid robots are far away from that.

We then looked at Airtasker as a marketplace platform. I think the moats in our business are potentially going to become even stronger during this period.

There is also Airtasker as an organisation, and how efficient we can make that organisation.

Although we provide our services via a piece of software, we don’t actually sell software. Airtasker is a net buyer of software.

As building software becomes faster and cheaper, we can deliver even more value to our customers without incurring as many costs. I think that factor is going to see a widening of margins in this kind of business.

I would differentiate us from a SaaS business that is selling the software.

The second thing we’ve observed in our business that’s exciting is that AI can help us use semantics and human-based language to get better at things like moderation in our community and to address platform leakage.

A lot of these things were hard to do with deterministic models. We’re trying to design algorithms to work out, for example, what is good behaviour and what’s bad behaviour.

Using generative models, we can more easily identify what, intuitively, is good marketplace behaviour that we want to encourage and what is marketplace behaviour we want to be able to address.

AI has really helped us to be able to do that.

The third thing is across our business, and so much of this is coming from the bottom up, which is exciting, but AI is already driving huge operational improvements.

For example, one aspect of Airtasker is it is difficult to identify customers at the precise moment at which they need a job done. Using agentic AI we can watch almost an unlimited amount of Instagram and TikTok reels and work out when a customer is in the market to buy a service.

By being able to stay on top of that in real time, in intuitive, semantic human language, we can meet those customers at that exact moment with an offer or a discount, and we’re seeing incredible conversion rates.

It is also helping us with aspects of our business, like marketing.

The agentic AI era is going to compound some important competitive advantages for Airtasker.

First are network effects. Ultimately this is a hugely powerful moat we are going to be able to build our business upon and that makes it very hard to disrupt.

The second thing is embedded transactions. Airtasker is running the payments. Being part of that economic value exchange, to disrupt the payments provides us with the opportunity to be able to hold people accountable and provide that transparency and accountability you need to run a powerful marketplace.

The third is proprietary reputation. People create their profiles on Airtasker, giving us unique inventory we can use to distribute the ability to buy local services through many of these agentic AI protocols and frameworks.

The fourth is regulatory responsibility. Airtasker provides a service to the Tax Office, for example, making sure that all our users are ID checked, provide tax information, etc.

This is generally something AI type models are probably not going to want to take on the responsibility for, and so this provides us with a huge competitive advantage.

Kinatico: A Change Management Factor

David Tasker: If we shift from marketplaces into regulated environments, the dynamic clearly changes. AI can clearly drive efficiency, but it also introduces accountability, risk and governance, and that’s exactly where Kinatico comes in.

What’s particularly interesting is their approach, embedding AI across the business while keeping humans firmly in control of decision making.

To take us through that, I’ll hand over to CEO Michael Ivanchenko.

Michael Ivanchenko: Kinatico is the involvement of PI, or personal information and personal data, across not only individuals, but corporates, etc.

Everything we do is about privacy by design, irrespective of technology, irrespective of approach, that is the primary consideration.

We’ve approached AI by first and foremost asking: how does it fit into our security frameworks?

How does it make sure we don’t violate any of the trust and confidence provided to us by our customers, and also make sure things like accountability, certainty and all of the things you have to have in an environment where you’re dealing in data systems, are actually prominent across the board?

What that also means is the adoption of AI in any company, but certainly in Kinatico, is a change management piece, just as much as it is a technology piece.

One of the reasons we appointed our chief AI officer, who is also our chief people officer, is to make sure the way we look at AI across the company is about accountable AI governance, including extending our existing ISO 27001 accreditation to the newly formed 42001, which is the International Management of AI.

We adopted Anthropic as our primary LLM within the organisation early in 2025 and adopted it across all parts of the organisation, to the point today where 88% of our employees are confidently utilising AI in their day-to-day work.

We’ve seen a 50% increase in features being deployed across all of our product set.

It is being used in all components of development, whether it is product management specifications, coding, of course, but then also QA, testing, and, most importantly, in any product company, the measurement of success, and iteration that feeds back into that product framework.

We’ve also developed our own proprietary LLMs to use within our product.

This isn’t about providing a tool set, it’s about ensuring you can deliver the value customers see and need at an increasing velocity, and on an increased basis generally.

AI is an opportunity to accelerate domain level expertise.

In our case, 17 years of domain expertise and proprietary data source access back to government that we use in all our verifications are not publicly available.

The data is certainly not searchable across the net, and is not something that governments, et cetera allow anybody to connect to because of data sovereignty and protection.

Everything we do is contained and maintained within our platforms. There is no feeding back into language models to help them learn, or other things like that. Everything we do is contained within our sphere.

Looking forward, this is a competitive advantage across marketing and sales. We want to make sure taking advantage of AI puts us in a position where we can start capturing the AI budget of organisations.

When you look at all the data that is coming out on how much companies are looking to spend and budget in their various departments, the one that is increasing is their automation across AI.

We want to make sure we are part of that ecosystem.

We’re not only embedded, but facilitating it and leveraging it, and being at the hub, if you will, of a hub and spoke type arrangement is a strong advantage for us.

From an operational point of view, the operational leverage of using AI for what it is particularly good at is in the automation and time reduction of certain processes, as long as you have the controls in place.

What is always important, and we’ve seen plenty of examples already in the press where AI deployments have gone awry, is that looking at the technology and the deployment of technology is the victory, rather than losing sight of the core KPIs, the reason you’re trying to do those things in the first place, and making sure that AI is delivering those advantages.

Using our core product as an example, Kinatico Compliance is built from the ground up with AI. It is a platform that is envisaged to be connected to by AI agents, not just people. Significantly, this includes even the pricing model we’ve been looking at.

We don’t charge for admin users, because that’s where the AI agents are going to be accessing the platform within constrained company environments.

What we charge is on the data generation, the value provided by monitoring and managing the compliance of the workers that are there.

We’ve also looked at overlaying it across our existing product space. An example of something we’re now rolling out has expected results across that resolution of customer inquiries, 60% automated, with a response time of less than two minutes versus potentially hours previously, and a cost per ticket reduction of around -40%.

One is the integration and access from AI agents, but the other one is allowing people to interact with computers in a way that is far more logical to themselves.

For instance, for a potential new starter, the platform takes over all of the things that need to be done, always with a confirmation, though. You notice it says: this is what I’m going to do, are you happy for me to proceed?

In summary, we’ve already embedded AI with over 80% of our staff using it daily. We think where we’ll get to is: on the physical ecosystem, we will end up with AI agents on our organisational charts.

All AI agents, and all people for that matter, are not infallible. Making sure the same safeguards we put in place today exist in an AI framework is also critically important, because, again, the benefits of the software, the benefits of AI technology, don’t negate the fundamentals of how you deal with your data security, your company security, across all of the workforce, your policy and obligations, your liabilities as a company.

Saying to somebody, yeah, look, that’s really bad, but it was our AI that did it, doesn’t get you away from the liability and accountability for you as an organisation and what you’re providing.

Understanding all of that, and the controls in place around it, become critical.

I think we will see productivity across organisations shift such that the size of organisations will not necessarily scale with more employees. Instead, companies will increasingly look at how their workforce can be supplemented with AI agents, particularly on the customer side.

We have seen this with every major technology evolution, whether it was the original internet or big data. Expectations around the speed of innovation, what customers can access, when and how they access it, and at what cost, continue to increase.

There is nothing fundamentally new here just because it is AI. The key issue remains ensuring companies deliver value and service, not just tool sets.

Any software company providing only tool sets risks being commoditised or replaced. This ties back to the core point that the fundamentals remain unchanged.

Organisations must focus on what makes customers want to engage and remain with them. At the same time, there is an opportunity to increase internal staff engagement.

This may seem counter-intuitive given the focus on workforce reductions, and while there are valid considerations around what that will look like, there is also a genuine opportunity.

For those who remain within organisations, AI tool sets and improved coordination can increase engagement. It is almost like giving every worker several personal assistants to help them perform their role.

Finally, within the technology ecosystem, the combination of proprietary large language models or AI models, alongside acquired models, proprietary and secure data, and AI-enabled processes and management, will be key to delivery.

This will be supported by appropriate interfaces that enable secure access and third-party integrations, forming the overarching technology model going forward.

Pureprofile: A Drive For Business Innovation

David Tasker: Our final presenter sits in a very different part of the AI landscape. This is a data and insights business where AI is not just enhancing workflows, it’s reshaping the product itself.

The key question becomes: how does AI drive both growth and margin expansion in a data driven business?

Martin Filz, CEO Pureprofile: Pureprofile is a data company and has nearly 1000 clients around the world who have business problems.

They come to us and we recruit millions of people around the world to be a source of truth. So that source of truth is behavioural data; what people do, what searches they’re doing, what websites they’re going to, etc, and clients can then analyse that data.

In the last few weeks, we’ve seen searches for electric vehicles go up over 50%. Our clients are looking to answer questions about what sort of automotive products people are interested in.

We have 14 offices around the world, about 260 staff. Nearly a third of our revenue is now platform revenue and about half our revenue is in long term annuity revenue; could be SaaS, could be long term contracts, again, AI is driving that as well.

We are a truly global company.

The first thing AI has been brilliant at is: it has driven innovation in companies. Because I can do it in AI doesn’t mean I’m necessarily going to do it in AI, but it means I start talking in the company more about innovation.

As an investor, you really want to see companies who aren’t necessarily just talking about AI, but are maybe accelerating their product development.

We’re all talking about AI. We may be writing birthday cards using AI, doing children’s homework using AI, teaching skills using AI.

The second thing that AI has done is the separation of coming up with an idea, which really product teams do, has changed.

What are clients asking for? What problems are we solving? Coming up with that idea, and then that gets thought out.

You’ve got ROI, you’ve got planning. Then that gets, if you like, thrown over the fence to an engineering team that says, OK, now go and build this piece.

And guess what? The engineering team goes back, looks at the specification, changes it maybe a little bit, and it always takes longer, and it always costs more money than you expected at the beginning.

Because of AI and because of this information, you’re seeing product design and engineering coming much closer together.

Prototypes that are coming out of product are 80% good enough, or, as we say, minimal viable products are 80% good enough to launch across the company, and quite often for internal tools.

What comes out of a product team allows you to start using something, maybe do some reiterations or iterations of that tool you’re using before you’ve got a final release.

Companies should be faster with new solutions.

I’m rolling out tools faster across my organisation, and then perhaps gaining efficiencies, because I’m not so heavy on the engineering side, expensive resources if they’re onshore (obviously less if they’re offshore), but I need fewer of those to do more work.

The third thing is everything is accelerated. Going from problem, design, prototype to launch is faster.

Keep an eye on companies’ employee satisfaction. Keep an eye on companies’ client NPS scores, satisfaction scores, because what you can see is, if organisations start to do too much too quickly, then they’re going to put a strain on their internal clients, and they’re going to put a strain on their external clients, and start to lose track of what they are actually in business to do.

What problem am I trying to solve? Really, I should just be using AI and technology to solve that problem faster and better.

I’m not trying to develop something newer. Sometimes technology can get in the way of what I do as a core business.

The speed that AI is now moving at is incredible. AI has gone from where it was really six to eight months ago, where it was aiding development, it was enabling development to be faster, to now, if I’m careful about it, AI can do all of the work for me, and I’ve got people overseeing it, but really all the code is being generated.

The problems are being solved just by AI. We’re up to version 5.3 of ChatGPT as an example. It doesn’t matter what AI you use, but in the latest release of 5.3, some of the code and product was totally developed by AI.

We now have this agentic AI, which is this terminology where, rather than using AI as an expert to speed up what I’m doing, you can actually say to AI, be the expert in the problem you are trying to solve.

We’re all using AI. We’re all using AI in our day-to-day life. People have downloaded it onto iPhones. Within companies you perhaps have safe, and you should have safe, enterprise versions, where you’re maybe uploading documents or asking business questions in a safe, ring-fenced environment.

Firstly, companies need to be more efficient. How can I replace manual tasks with automation and do more with less? However, and this is a huge however, you are, just as a company, changing your risk profile.

At our core, we have irreplaceable data that is updated daily by millions of people around the world. Technology is no longer a moat. You need to look at companies and ask: what are their moats? What are their defensible points of business?

Because we’re seeing companies like Atlassian, where their share price is being decimated; or are there other reasons for that?

Is it replaceable?

For a company like WiseTech Global ((WTC)) the defensive moat is the relationships and how they’re embedded across the whole shipment plan, the delivery, the route to market, from product being put onto tankers and delivered to end case, that’s the defensible part, not the technology.

Watch AI washing. What I do as a company, Pureprofile, is answer business questions that companies have; they want to understand more from their consumers.

I’ve not suddenly become an AI company. I just use AI to do that better.