Making AI Practical: From Strategy to Execution
In our previous article, we explored what the primary characteristics of a successful AI-first operating model look like. Today, we shift from diagnosis to application. If the challenge is translating AI potential into operational practice, the solution lies in understanding where and how AI creates value across the business. We begin with member engagement and customer experience: the front line where AI’s impact is most visible to the people your organisation exists to serve.
And a bit of a drafting note – at Kyudo, we serve clients across government, financial services and the superannuation sector. In this series we provide examples from the superannuation sector, but these can be easily transposed into your specific operations, regardless of where you spend your time.
Let’s dig in…….
Current State Challenges
Financial services institutions struggle to deliver personalised experiences at scale. Segmentation models group millions of individuals into demographic cohorts rather than understanding unique needs. Generic communications fail to resonate. Response times lag customer expectations set by retail banking and digital platforms.
In superannuation, the challenge intensifies. Members engage infrequently, often only during moments of crisis or major life changes. When they do engage, they expect sophisticated service instantly. Traditional engagement models, built for batch processing and annual statements, can’t meet this demand.
According to HubSpot’s 2025 State of Customer Service research, customer service teams using AI have cut call handling time by 45% and resolved issues 44% faster, with 92% of service leaders reporting improved response times. The question isn’t whether AI can help, but how to deploy it without creating new problems.
AI Applications
Personalised Advice at Scale
SuperAnne, an Australian conversational AI platform for superannuation funds, demonstrates what becomes possible when natural language processing meets financial guidance. The platform speaks to members in plain language, models personalised scenarios in real time, and offers practical guidance for decisions that shape financial futures. It bridges accessibility, affordability and trust in ways that human advisers alone cannot achieve at scale.
The technology works because it combines several AI capabilities: natural language understanding to interpret member questions, scenario modelling to project outcomes, and conversational design to maintain engagement. None of these capabilities is revolutionary individually. Combined thoughtfully, they change what’s possible.
As fun experiment, here is an AI generated video of my older self giving my 25 year old self a ‘bite-sized’ piece of personal advice. The experiment contemplates a technical solution that uses open banking capability to tailor the advice for this specific individual – a true ‘category of one’.
Intelligent Routing and Case Management
AI-powered routing systems analyse the complexity, urgency and context of member enquiries to direct them to the most appropriate resource. In superannuation, this might mean automatically identifying a member experiencing financial hardship based on transaction patterns and proactively connecting them with support services before they request assistance.
More than 70% of customers expect companies to cater to their needs personally, and failure to do so damages brand loyalty. Intelligent routing isn’t just about efficiency. It’s about ensuring members reach the right help at the right time.
Proactive Communications
Rather than batch-and-blast campaigns, AI enables event-driven, contextually relevant outreach. For superannuation funds, this means detecting a member who has stopped regular transactions despite strong balances and offering targeted support rather than generic campaigns. AI enables better segmentation than traditional demographic approaches, using unsupervised machine learning to automatically identify meaningful patterns in member behaviour.
Cross-Sector Parallel: Government Services
Municipalities are deploying AI chatbots, SMS alerts and mobile apps to give residents instant access to information and services. The patterns transfer directly to superannuation: high-volume, low-complexity interactions automated; complex, high-stakes matters escalated to specialists; satisfaction improving despite reduced direct human contact for routine matters.
Operating Model Implications
Implementing AI for member engagement restructures how customer-facing teams operate. New roles emerge: AI Product Owners sitting between technology and member services, Conversation Designers building natural language interfaces, and data scientists embedded in frontline operations. The shift from campaign-based to always-on, event-driven engagement requires different processes and real-time decisioning infrastructure.
Frontline staff must learn to work alongside AI rather than compete with it, requiring prompt engineering skills and AI literacy to interpret model outputs. Data infrastructure demands unified member platforms breaking down silos, real-time pipelines replacing batch processing, and privacy-preserving techniques for personalisation.
Implementation Considerations
Start with Process Mapping, Not Technology
Before selecting AI tools, map your current member journey end-to-end. Identify where members abandon interactions, where repeat contacts occur, and where staff spend time on avoidable questions.
For member services teams, this means tracking metrics for a period to collect data (say, 30 days). These metrics may be things like question categories (what are members actually asking?), resolution paths (how many touches to resolve?), and escalation patterns (what triggers human involvement?). This data determines which AI applications will deliver genuine value versus creating new work.
Build the Escalation Framework First
The most common implementation failure is deploying AI without clear escalation protocols. Define precisely when AI hands off to humans: complexity thresholds, emotional distress signals, regulatory triggers, and member preference indicators. Train the AI and the team simultaneously on these handoff rules.
Your frontline staff need to know their role isn’t being eliminated but elevated. They become exception handlers, complex case specialists, and AI quality controllers. Make this explicit in job descriptions, performance metrics, and training programmes.
Pilot with Willing Participants Who Understand the Context
Invite members to opt in to the AI pilot with clear communication: “We’re testing new AI tools to improve service. You’ll get faster responses, but there may be occasional bumps. You can opt out anytime and speak with our team instead.”
This approach delivers three benefits: participants are more forgiving of early-stage limitations, they provide better feedback because they understand they’re part of development, and you avoid the regulatory and reputational risk of testing unproven tools on vulnerable members or complex cases. If you want to be bold, consider offering incentives for those who help refine your processes and tools.
Target 200-500 pilot participants across different demographics and engagement levels. Run parallel operations for 90 days: AI handles enquiries with human review of every tenth interaction. Track where AI misunderstands context, provides inappropriate responses, or misses member intent. Use these insights to refine before broader rollout.
Keep vulnerable member segments—financial hardship, recent bereaved, disputes—on traditional service channels until AI proves reliable in lower-risk scenarios.
Measure What Frontline Teams Care About
Traditional metrics—cost per contact, handle time—miss what matters. Measure instead: member effort score (how hard was it to get help?), first-contact resolution rate, staff confidence in AI outputs, and time freed for complex cases.
Quick wins validate the approach: FAQ chatbots show results in 8-12 weeks, sentiment analysis in member emails surfaces issues within days, and intelligent routing reduces misdirected enquiries by 40-50% in the first month.
Regulatory Compliance as Design Input, Not Afterthought
Any AI interaction touching financial advice must meet best interests duty requirements from day one. This means every AI recommendation includes explainability (“here’s why I’m suggesting this”), audit trails for regulator review, and bias testing across age, gender, balance and engagement segments. Build regulatory evidence collection into the AI workflow, not as a retrospective compliance exercise.
Next in this series: Next, we examine how AI transforms operations and administration—moving from manual processing to intelligent automation that learns and improves.
Selected References
- HubSpot (2025). 70+ Customer Service Statistics to Know in 2025 (+ State of Customer Service data). Published August 7, 2025.
- Australian FinTech (2025). SuperAnne: The AI Platform Bringing Human Connection Back to Superannuation.
- McKinsey & Company (2025). Customer Personalization Research. Cited in: CMSwire (August 27, 2025). “CX Personalization Statistics in the US for 2025
