How Brokers Can Use AI and Automations Without Overcomplicating Their Tech Stack — with Simone Blakers from Liquid CX

1. From Data-Driven Marketing to Digital Workforces

Simone started her career roughly 30 years ago as a data-driven marketer, back when direct marketing was king. She followed the industry through the dotcom boom into digital, spent years building databases, and worked heavily in predictive modelling as it emerged. After leading globally awarded agencies across the digital, data and customer experience space, she went independent eight years ago with Liquid CX — a consultancy focused on helping clients apply data and technology to solve business problems, from marketing all the way through to process automation.

The through-line of the whole conversation is a mindset shift: moving from thinking about tools and systems in your business to thinking about digital workers — a digital workforce that actually takes action on your behalf.

2. Why Data Is the Foundation of Everything

Before any conversation about AI, Simone grounds it in data. Every prospective, current and past client represents a body of data — from basic personally identifiable information like name and email, through to the more sensitive information brokers hold once a client is in an active broking relationship and identity checks come into play.

The core skill is knowing how you hold that data, where you hold it, and how you activate it in the right way. A CRM sits at the centre of this as the foundational system that holds client information and enables client-facing teams to use it. The evolution Simone points to is that a "client-facing team" may increasingly include a digital worker — an AI agent — alongside human staff.

3. The Four Levels: Automation vs AI Automation vs AI Chat vs Autonomous Agent

This is the heart of the episode. Simone and break the landscape into distinct categories that are often lumped together and confused:

• Basic automation. A simple sequence of events you've wired together manually — a calendar booking flowing into your CRM and triggering a confirmation email. No AI required at all.
• AI automation. The same kind of workflow, but with a large language model playing a role at some point in the sequence. The AI part is typically what makes something conversational or personalised per person. Instead of a fixed template ("Hi [name]"), the AI can pull data from your CRM and adapt: "I noticed you prefer to meet for a coffee at such-and-such — shall we?"
• AI chat. A chat interface like ChatGPT, Copilot, Claude, Gemini or Perplexity. It responds to the input you give it and returns a response. Even a custom GPT grounded in your business knowledge and tone of voice is still not an agent — because it isn't taking action.
• Autonomous agent (a digital worker). Here you give the AI agency — authority to take action on the business's behalf, much like an employee. Instead of you sequencing the steps, it figures out the steps itself to reach an outcome you've defined.

Simone illustrates the top end with a scheduling example. Ask an autonomous agent to arrange a client meeting, and for one client it might search locations, find a reasonable café near their office, make the booking, send calendar invites to both parties, and text both a reminder with a map link on the morning. For another client who lives near the office and prefers to come in, the same agent might instead phone the client directly to align diaries — different behaviour, same agent, no pre-set sequence.

4. Where AI Fits Into a Real Broker Workflow

Walking through a practical front-to-back flow, the natural first place to introduce AI is meeting notes. Rather than writing up notes manually after a call, an AI can listen, transcribe, summarise, and push notes back into the CRM.

But Simone flags an important control point: that summary feeds everything downstream. Because the CRM is your source of truth, if the AI-generated notes are wrong, every follow-up email built on them will be wrong too. This is why a human approval step — having the note checked by the person who took the call before it's published — is often worth building in. Bad data means bad information, all the way down the chain.

5. Computer-Using Agents: Powerful, but Early Days

Simone describes the emerging frontier: computer-using agents that open a desktop environment and operate a mouse to complete tasks across different websites — filling in different quote-request forms for local deck builders, for example, without being told the exact steps for each site.

Her honest assessment for brokers: these tools exist now and are worth playing with, but applying them in a business context — particularly one handling sensitive client data — is still very early days. The advice is to start simple, apply basic automation where it has immediate impact, and only then move to safe, low-risk AI within an agentic flow.

6. The Right Way to Start: Design the Experience First

Simone is firm that automation should never be the starting point. The sequence she recommends is:

1. Strategic experience design. Decide what you want clients to feel and experience. Identify the moments that matter for genuine human connection and relationship building, versus the moments that can be automated. If you can define the experience for your people, you can define it for automation and agents.
2. Opportunity and pain-point analysis. Look at where the team gets bogged down — repetitive tasks, paperwork, slow processes — and where clients feel friction. Often clients who haven't bought or sold in ten years need more handholding and education than a busy team has time to give; that's exactly where automation and AI support can lift the experience.
3. Risk analysis. Ask where it would be genuinely devastating if something went wrong — identity verification being an obvious high-risk zone — and be deliberate about where you're willing to hand authority to an automation or AI.

A frictionless process reduces effort for the client and the broker at the same time, which is where the big wins come from.

7. The Risk and Data Privacy Layer

Risk isn't black and white — it's a gradient that shifts depending on what's connected, how it's connected, and how it's used. Simone highlights several considerations brokers cannot skip:

• Data privacy. Even when simply using ChatGPT, Copilot, Gemini or Claude, make sure model training is switched off so you're not training the underlying model on client data, and confirm there's no data leakage — in line with the commitments in your own privacy statements.
• Data protection and retention. Once you've captured data, you need to protect it: know how long you're keeping it, what your deletion protocols are, and whether you have the right security in place.
• Cyber security. Introducing AI can introduce a new layer of cyber security exposure.
• The redaction trap. Even with tools that redact sensitive documents (such as removing a tax file number), there's a vulnerable moment: the document has to be held somewhere before the redaction happens, and that holding point is where the exposure sits. This matters especially in legal contexts, but brokers handle plenty of sensitive documents too.
• Breach readiness. Under the data breach notification scheme, every business holding sensitive information should have a scenario plan and a complete response plan ready before a breach happens.

8. AI Literacy Comes Before AI Strategy

One of Simone's strongest points: before you can even discuss risk meaningfully, the team needs a baseline of AI literacy — understanding what AI is good at, what it's not, how it works, and where its vulnerabilities lie.

The inherent risks exist even for internal, non-client-facing use: bias, accidentally including personally identifiable information, data leakage, and hallucination (confidently incorrect answers). Simone finds that teams often need to build this base literacy first, because otherwise conversations about strategic use get stuck explaining how the technology works rather than how to manage its risks. Her recommendation is to raise literacy across the team, then bring in people who can help identify where the specific risks sit.

9. Choosing the Right Tools for Your Stack

Simone touches on how the tech stack should be constructed — often with the application layer hosted separately from the CRM that holds your data. The platform landscape she references includes:

• Salesforce with Agentforce built in
• Microsoft Copilot Studio for those in the Microsoft environment
• Relevance AI — an Australian success story focused on building a digital workforce
• Enterprise-grade robotic process automation such as Blue Prism, Pega and ServiceNow

Katey adds a preference for Botpress for self-hosted, low-cost bots where you can store your own information in a protected environment — distinct from marketing-oriented tools like ManyChat. Many of these platforms are now low-code, letting a business user who understands the process build the solution — though Simone stresses you still need someone with a technical eye across integrations and security vulnerabilities.

10. The Bigger Opportunity: Superpowers for Small Teams

Beyond efficiency, Simone points to how AI is putting capabilities that were once enterprise-only within reach of small and medium businesses. Basic CRM systems can now track behavioural signals — email opens, site visits, a client warming up again — and prompt a timely, human conversation. AI can help build small mini-apps that would previously have taken far longer to develop.

The standout use case for brokers: because you know your client better than anyone and hold great structured data, the right AI agent could, for example, begin identifying investment-suitable properties for a client flagged as an investor personality — expanding a broker's service into property finding, something that would once have felt out of reach. As Simone puts it, AI is helping teams do things that used to be beyond them.

11. Structured vs Unstructured Data — Why the CRM Still Matters

A technical but important closing point: large language models are excellent with unstructured data — conversations, images, voice. But activating agents, automations and workflows still depends on structured data. Knowing that the Simone who emailed from a personal Gmail is the same Simone from a business address, that she has multiple properties, or that she previously lived at one address and is buying another — all of that structured, connected data is what makes agents reliable. This is precisely why the CRM remains foundational and can't be skipped.

12. Key Takeaways for Brokers

• Start with the experience, not the tool. Define what you want clients to feel first, then decide where automation and AI belong.
• Know the four levels. Basic automation, AI automation, AI chat and autonomous agents are different things with different risk profiles. Don't conflate them.
• Protect your source of truth. The CRM holds the truth — build human check-points so AI-generated notes don't corrupt everything downstream.
• Literacy before strategy. Raise your team's baseline AI literacy before rolling AI into client-facing processes.
• Treat risk as a gradient. Start with low-risk applications, switch off model training, know your data retention and deletion protocols, and have a breach response plan ready.
• Start simple. Basic automation alone can have a big impact. Computer-using agents are exciting but still early days for sensitive client work.

13. Practical Next Steps

1. Map your current client journey and mark the moments that must stay human versus the ones that could be automated.
2. Identify your team's biggest repetitive pain points — these are your first automation candidates.
3. Audit your existing AI use: is model training switched off on every tool touching client data?
4. Build a human approval step into any AI that writes notes or drafts client communications.
5. Assess your team's AI literacy honestly, and consider foundational training before scaling AI use. Free starting points include Google's courses on Coursera and TAFE's AI foundations.
6. Draft (or dust off) your data breach response plan.

Episode Links
▶️ WATCH THE FULL PODCAST HERE: https://youtu.be/WI06zaNJPes

🎧 LISTEN ON SPOTIFY: https://spotifycreators-web.app.link/e/ZFBFHjJby4b

🌐 LiquidCX.com.au — Simone's consultancy applying data and technology to solve business problems

👤 Connect with Simone: LinkedIn