Navigating a new development model

Interactive Voice Response (IVR) solutions once frustrated both callers and developers with repeated “sorry, I didn’t understand” messages. In traditional software development, you wrote code to send an API request, parse a defined response and handle known errors in predictable ways. Testing was clear-cut.

Today, with LLM-based agents, development works differently. Developers guide models to say “I don’t know” when appropriate – an essential part of building trustworthy experiences. This requires prompting the model to reason probabilistically, rather than following fixed rules.

The shift means guiding the model to associate customer intent with the correct API call and the necessary parameters to execute the API call successfully – a more flexible, less deterministic approach.

Designing for consistency and clarity

Even small changes in phrasing or context can lead to variations in the model’s interpretation. As a result, evaluating performance is no longer about binary success. Instead, it’s about assessing how reliably the model maps natural language to structured API calls across a wide variety of inputs, edge cases and unseen user behaviour.

With new metrics and processes required, bridging the gap from proof-of-concept to production-grade application becomes a significant business challenge. Indeed, Gartner predicts that at least 30% of generative AI projects will be abandoned after proof-of-concept by the end of 2025. As AI technology advances, aligning enterprise expectations, workflows and development methods will be critical to delivering long-term value.

Voice AI’s transformative promise

Even with these new dynamics, the opportunity for voice AI agents is vast. Contact centres – complex, high-volume environments – may be transformed by agentic solutions that can scale, support human agents or enable autonomous customer experiences.

For the first time, voice becomes a truly intelligent interface. AI agents can interact conversationally, adapt in real time, and generate new insights into customer preferences and the overall customer journey at a previously unimaginable scale.

This isn’t just a new channel – it’s a smarter way to support one of the enterprise’s most personal customer touchpoints.

Building for production: PolyAI’s approach

PolyAI has worked with LLMs since 2017, partnering with leading enterprises like Pacific Gas & Electric, Caesars Entertainment and Unicredit to deploy voice agents that manage millions of conversations with fluency and enterprise-grade reliability.

That experience shaped PolyAI Agent Studio – our enterprise platform designed to help teams build, manage and continuously improve production-grade voice agents.

Powering our platform is a tightly integrated suite of proprietary models: Owl for speech recognition and Raven for reasoning. These models give us something off-the-shelf systems can’t – fine-grained control, deep observability and continuous learning from real-world conversations.

With reinforcement fine-tuning, our agents become more reliable over time, improving how they respond, manage uncertainty and support seamless customer conversations.

Our context-orchestration framework lets developers connect from CRMs, telephony, APIs or user history with precision, allowing personalised, brand-consistent interactions at scale while limiting the risk of hallucinations.

At the heart of our platform is fluency: enabling AI agents to respond with relevance, continuity and clarity.

Deploying with Microsoft Azure

By extending Agent Studio to Microsoft Azure, enterprises gain more control over how they deploy PolyAI agents, whether for compliance, data residency or closer integration with their Microsoft ecosystem.

We’re also building integrations with Dynamics 365 Contact Center and Microsoft Teams, enabling faster deployment of voice AI agents into high-value customer service workflows.

Find out more

Visit the Microsoft Azure Marketplace for more about PolyAI’s solution

Read MIT’s Technology Review Insights Report on Customizing Generative AI

About the author

Michael Chen is the VP of Strategic Alliances and Corporate Development at PolyAI. He leads a team focused on expanding PolyAI’s collaborative relationships with cloud providers, technology partners and global systems integrators. Michael has been on the frontier of the LLM and generative AI revolution since 2020, bringing expertise and perspectives across corporate strategy, customer experience, product marketing and business transformation.

The post How PolyAI helps enterprises deploy production-grade voice AI agents faster appeared first on Microsoft Industry Blogs - United Kingdom.

The emergence of AI agents is poised to further accelerate innovation. In this article, we’ll explore the impact of these agents on the insurance sector and how organisations are using them to transform business operations.   

What are AI-powered agents?  

Often referred to as “orchestrator agents”, autonomous AI agents are designed to achieve specific goals with reduced intervention from us. Extracting greater value from existing data, reducing operational costs, automating processes and improving decision-making. They can also learn from past interactions and even execute tasks autonomously.

While these improvements are promising, safety is key to rolling out agentic systems. This means prioritising security, privacy and safety throughout AI development. Drawing on decades of expertise in building AI products, Microsoft is committed to industry-leading practices and capabilities. This includes our Secure Future Initiative and Responsible AI principles, providing a solid foundation to unlock AI’s transformative potential with confidence.  

These efforts enable insurers to enhance the customer experience while maintaining outstanding standards of trust and reliability.

Using agents in the insurance industry 

Here are some significant industry use cases. 

Transforming underwriting through time-saving

AI-enabled tools in underwriting can automate data ingestion, translate unstructured data into structured data, and help underwriters access and interpret information from multiple, isolated sources.  

For example, Markerstudy Group built an app for its claims department on Azure OpenAI to summarise claims calls for handlers. With 300 claims handlers using the app, the organisation saved around four minutes per call – and a total of 7,500 working days.

AI agents can also provide real-time cross-checks, ensure compliance with new regulations, and help underwriters evaluate more policies by removing manual tasks from their workflow. For example, Zurich Insurance Group is using tools like Azure OpenAI to support more accurate, efficient risk evaluations – accelerating underwriting and boosting customer satisfaction.

Automating claims processes and data analysis

Agentic AI can automate the extraction and validation of claims data from forms, emails or portals. It can also cross-reference claim information with policies and coverage terms, and route claims for approval or escalation based on pre-defined rules.  

Insurance platform Nsure.com has created a customer copilot which draws data from Microsoft Fabric and helps customers submit payments, review renewal offers and request discounts. Along with text, the copilot offers an interactive voice response option and handles around 60% of customer questions. In scenarios like this, agents can reduce the claims process from weeks to hours.  

Improving customer services and empowering employees

With chatbots available 24/7 and capable of understanding natural language, AI agents can increase customer choice, personalisation and satisfaction. 

When a customer initiates a claim, they typically complete a form to gather general information and open the damage case. To streamline this process and improve service quality, insurance company DOMCURA has developed an AI-powered agent called Claimens, capable of handling calls, guiding customers through the claim form and outlining next steps.  

By handling simple tasks, the agent enables case workers to focus on more complex claims, which helps improve the experience for both customers and employees. 

Take the next step with trusted support 

With the UK government’s Action Plan emphasising the role of AI in driving economic growth, implementing AI-powered agents is now a key strategic priority and differentiator for businesses.  

The value of agents comes from their potential to support complex use cases across business functions, particularly for workflows that involve time-consuming tasks or require multiple sources of data. To get started with Microsoft as your partner, we invite you to co-innovate AI use cases to ensure we’re transforming priority areas.  

By leveraging proven solutions, you can streamline processes and improve customer satisfaction while maintaining the highest standards of trust and reliability. 

To learn how to build your strategic plan today and discover the five stages of AI value creation, download a free copy of our AI Strategy Roadmap.  

Find out more

Read the blog: Insurance in the era of generative AI – use cases that transform the industry

About the author

James has held leadership positions across Microsoft Consumer and Enterprise teams. With a passion for people and technology coming together to drive customer success, he has been at the forefront of cloud and digital transformation for over a decade, developing new revenue streams and significantly accelerating growth.

The post Transforming insurance with AI-powered autonomous agents  appeared first on Microsoft Industry Blogs - United Kingdom.

Today their role hasn’t necessarily changed. However, the value organisations place on them certainly has. The focus is shifting from fitting customers around business processes to reshaping contact centres around customers’ needs.

For years, the role of contact centres was limited – often confined by traditional 9-5 working hours. It was predominantly aimed at driving down costs and improving efficiencies.

This was reflected by the way companies measured their success. They had KPIs ranging from targets for call volumes to queue times and abandonment rates. These inward-focussed efficiency metrics have, however, consistently failed to put the customer at the centre of the service.

In today’s increasingly digitalised environment, this is no longer sustainable. Nothing is more valuable than customer experience and customer outcome. Organisations are fast adapting to the idea that great customer experiences convert into customer loyalty and new customers. People increasingly sharing their positive and negative experiences online. As a result, financial institutions can no longer afford to underestimate their services.

Contact centres are transforming. From unempathetic, 9-5 services reliant on a standard agent script, to becoming a customer experience centre. They don’t just focus on a service but the total customer experience across an organisation.

This presents a new opportunity for financial services companies to become fully connected organisations driven by technology. Embrace solutions that connect and unify all their channels – from digital to physical and mobile. As a result, they can create seamless, connected customer experiences that distinguish them from their competitors.

Understanding the needs of financial services customers

To better equip contact centres to service customers, we first need to look at how the needs of these customers have changed over time.

The past few years have seen the customer landscape evolve and diversify significantly. Alongside more traditional customers, organisations are increasingly welcoming a new generation of tech-savvy, socially connected customers. They come with a fresh new range of expectations.

Empathy, passion and hyper-personal connections are key drivers behind their demands. They centre around being understood and supported throughout their customer journey. Failure to do so can have catastrophic effects for organisations. Not only will it risk customers leaving their service but also expressing their frustration online.

This means one thing:

The more you know your customer, the more you can tailor your service to them.

A customer who’s been with your organisation for decades will be likely to seek support through traditional landlines or your website. On the other hand, the younger, digitally savvy customers will want mobile and self-service options, pursuing a more digital experience.

So how can organisations make sure that all these needs and preferences are satisfied? Put simply, the more diversified the audience, the more diversified the services.

Breaking down silos in contact centres

To really drive customer satisfaction across your evolving customer base, you need to invest in omnichannel engagement. Encompassing anything from social media to instant messaging, webchats and physical customer support, customers choose their channel of preference.

But this hasn’t always been the case for organisations in the financial services industry. Organisations may have invested in technologies to support a growing number and type of customer-facing channels. However, these are often used in silos and operated by different vendors.

This leaves customer data confined. Additionally, it prevents agents from surfacing customers across multiple systems. Most importantly, it prevents organisations from leveraging customer insights and using them to better orchestrate the customer journey.

Organisations who adapt and unify these siloes will be more likely to succeed at improving the customer journey. Doing so will empower employees to be more collaborative and productive. It will also reduce time to serve customers and provide an overall higher quality of service.

But it’s not enough to change the internal ways of working. Organisations must improve the way they build relationships with their customers. Looking ahead, they need to improve their ability to capture interactions in the moments that matter. They must continuously adapt and improve using this new-found knowledge.

To do this, they need an infrastructure and technology foundation. One that can empower them to capture these moments, understand their context and orchestrate the best, most optimal route across any function. All to deliver fast, impactful and personalised services that convert prospects into long-lasting advocates.

The rise in automated self-service technology

In a world that increasingly relies on digital innovation and newly found tech capabilities, automation can play a key role in improving customer services and contact centres.

Until recently, these have had virtually no front-door filter standing between customers and operators. Self-service has only just started to become a reality, leaving agents to deal with more complex cases.

This is where automation comes in. As data-based insights and capabilities become the norm, organisations have the opportunity to identify the simpler customer queries. They can then direct them to self-service areas, virtual assistants and AI-powered services.

Conversational virtual assistants are a powerful tool. Especially when it comes to harnessing data to gain insights on the customer. This data can be used to understand customer demands, their purchase history and previous complaints and other crucial information that can help them address their query entirely autonomously.

If the customer wants to transfer to a human, all that data can be carried across. Using AI, potential knowledge articles and recommendations, agents can successfully solve a customer’s request.

AI can also assist with more complex tasks such as pre-authenticating customers before speaking to an agent. This time-saving feature benefits both the customer experience and a contact centre’s inward metrics. With the addition of voice-biometric technology, a virtual agent could also help detect and prevent fraud by comparing a customer’s voice against their customer profile. A more cost-effective solution to training agents on fraud prevention and extra reassurance to customers that their money is secure.

These kinds of innovations aren’t there to make calling a contact centre redundant. There will always be a need to speak to agents to help manage banking relationships or advise on future monetary decisions. But for simpler, everyday tasks, financial organisations can empower customers to self-service rather than waiting to speak to an adviser.

Challenger banks have been particularly good at pushing innovations in this way and raising the customer service bar. Many of them are truly revolutionising retail banking by reducing typical applications processes from a week to minutes. By promoting a digitally-native experience, more traditional banks are forced to reconsider their own customer experience.

Keeping customer data secure in the cloud

Data breaches happen far too frequently today. And as financial institutions can hold an entire customer’s wealth – from mortgages to loans to bank balances – there’s an enormous responsibility to ensure that data is kept safe and secure.

This presents an immediate challenge to spend millions innovating on an existing IT infrastructure. This may require a huge amount of capital investment and resources to maintain. We’re seeing many leading insurance companies and banks choosing to migrate their contact centre operations from on-premise servers to the cloud.

If you consider Azure for example, Microsoft has already spent billions creating a secure cloud solution and helped protect leading organisations from cyber-attacks, fraud and Denial-of-Service on an intraday basis. This reassurance makes migrating to the cloud not just a business decision for better data security, but also for greater cost efficiency by eliminating the many overheads that physical servers require.

The cloud also offers advantages when it comes to complying to financial regulations such as how organisations handle data, offer services and prevent financial crime. By working with a trusted cloud provider like Microsoft, a lot of this responsibly can be shared and evidence can be provided to show that data is being kept securely and systems are operating within regulations.

An all-in-one solution for financial services contact centres

Financial organisations are changing. Their reputation and global presence is increasingly tied to customer experience, online reviews and the quality of their services. As a result, they must reimagine their services with a new, more demanding and diversified customer base in mind.

At the same time, switching banks or insurers has never been simpler. Therefore, it crucial for organisations to innovate their contact centre and make the end-to-end experience as efficient and helpful as possible.

The key is to not consider every channel as a separate challenge. A 2021 Forrester report commissioned by Microsoft, Boost Your CX With A Better Integrated Contact Center, CRM, And Collaboration Systems, found that 74 percent of contact centre agents in organisations typically use four or more applications to service customers. This gives a disconnected experience for agents. But by implementing an all-in-one contact centre solution such as Microsoft Dynamics 365 Customer Service, financial organisations can manage their operation through a single platform. From initial customer contact to automated self-service with AI virtual assistants, to agent-guided case management and back office collaboration with Microsoft Teams.

This allows live agents to interact with customers on any channel. They have a complete overview of all previous interactions to give a frictionless and effective customer journey. It also helps to free up their time. So they can focus on the most complex and sensitive requests that virtual assistants aren’t equipped to handle.

Find out more

Envisioning the Future of Customer Experience

Microsoft Dynamics 365 Customer Service

About the author

Chris leads the Dynamics 365 Customer Engagement portfolio for Microsoft UK within the Dynamics 365 Business Group. Chris is responsible for developing and orchestrating the go-to-market strategy across this portfolio for the UK geography to generate awareness, create excitement and drive business development. The Dynamics 365 Customer Engagement portfolio is a suite of intelligent front office business applications designed to accelerate digital transformation across sales, marketing and customer service.

The post How data and AI will transform contact centres for financial services appeared first on Microsoft Industry Blogs - United Kingdom.

Following our simulated hospital event at UCLan in May 2020, I built a selection of chatbots using Power Apps. These were developed with the primary purpose of increasing fidelity within online simulation for our health and social care students. I wanted to explore how to automate and standardise parts of the simulation, to help with quality and control from the facilitator’s perspective.

Creating and developing a chatbot

To create a chatbot, I used Power Apps. I first selected Create under the left hand Chatbots heading, and then worked on developing the language behind the bot.

To get started I’d recommend using Microsoft Whiteboard or some paper to map out the language according to what you want the bot to ask and respond with. Try to draw on the direction of the conversation, and how you want it to start and end. Spend some time exploring existing tutorials found in Power Apps, Power Virtual Agents, Azure, and LinkedIn Learning.

If you’re not sure where the bot will be deployed just yet, consider starting in Power Apps. This will make it possible to save it and then ‘sideload’ into a Team, or to embed the bot into another web site or resource. Follow the Microsoft Power Apps Community for Q&A and problem solving tips to help you along the way. 

Once created, you can proceed to Publish the chatbot, following any further instructions.

Taking a student-centric approach

It’s also important to consider the different types of chatbot that can be created when looking to develop your own. In our case, this helped shape how the bots would be used and where they would ultimately be deployed. Here are three different types of chatbots and examples of how we have used them at UCLan.

1- Repetitive prompter chatbot

Within one online OSCE for our MSc Occupational Therapy students, there are four tasks. The elements of each task became the most asked question, both across the module and leading up to the OSCE. Students understood the tasks but often couldn’t retain which order they were in, and whether these were live or recorded. I created a ‘repetitive prompter’ style chatbot for the module, and we have had real success embedding and deploying it within the assessment space. This is quite an exciting development for me, as it reinforces the benefits of investing in time in chatbots – particularly due to the potential of them being embedded within other Virtual Learning Environments. You can also lift this type of chatbot and place it within your Microsoft Teams space.

2- Reflective prompter chatbot

Our debrief chatbot for IPE online simulation was a collaboration between my colleague Abhi and I. Abhi came up with the language for the chatbot, and I then programmed and published it in Power Apps, using the demo link as a Website tab in Microsoft Teams. This was for a large-scale simulation, with 300+ people within three Teams environments on the one day. This was the first cross-faculty IPE event which had been run online whilst students were studying from home. Subsequent feedback from students and staff has since provided valuable insight into further development for future events. With growing interest in online simulation, we created a Team purely for staff to collaborate around simulation ideas. This has proved really useful, as the bots are now becoming shareable assets across our wider organisation.

3 – Role-specific chatbot

Development of role-specific chatbots started from reflections about how to part-automate online simulation, and also the possibility of deploying a series of chatbots within an online simulation to work as simulated characters within the Microsoft Teams environment. I created a couple of ‘patient’ chatbots, including ‘George’- who needs the toilet. If you don’t respond appropriately, he (understandably) becomes very angry, upset and intends on placing a complaint. I am currently developing another simulated hospital event for 2021 and intend for this to be an IPE activity.

Here is an example of a nurse chatbot which we deployed into a large-scale IPE simulation for a colleague.

Looking ahead

Having successfully deployed chatbots at a local level, we are now looking at the possibility of wider publishing to the organisations app catalogue in Microsoft Teams. Bots could then be selected by other users, and once deployed, would appear as 1:1 chats.

We have no doubt that 2021 and beyond will highlight an increased appetite in the part-automation of a number of day-to-day tasks undertaken by educators, and this is something we will continue to both explore and evaluate.

Find out more

Chatbots for TEL

Creating chatbots for online simulation

Power Apps

App Studio: Creating chatbots in Microsoft Teams

Power Virtual Agents

Learning Microsoft Power Apps

Read more education blogs 

About the author

Sam Pywell is a Lecturer in Occupational Therapy at the University of Central Lancashire, and an Associate Fellow of the Higher Education Academy. Sam has recently led development of chatbots in online simulation for health and social care students using Microsoft Teams. She is an MIE Expert, DigiLearn Champion and Key Contributor to the DigiLearn Sector community. You can follow her on Twitter @smileyfacehalo.

The post How to develop a chatbot to support your educators and students appeared first on Microsoft Industry Blogs - United Kingdom.

[msce_cta layout=”image_right” align=”right” linktype=”blue” imageurl=”http://approjects.co.za/?big=en-us/industry/blog/wp-content/uploads/sites/22/2020/09/Pillar-5-banking.png” linkurl=”https://info.microsoft.com/UK-DTFIN-CNTNT-FY21-10Oct-14-Boostingtheinnovationofbankingbusinessmodels-AID-3024293-SRGCM3895_01Registration-ForminBody.html” linkscreenreadertext=”Download the report” linktext=”Download the report” imageid=”46091″ ][/msce_cta]

The digital customer experience has never been more important. Consumers are more mindful about world problems such as global warming and others. All of this translates into expecting their banking provider to act as responsible corporate citizens while offering advanced digital experiences.

How technology can help drive innovation

 Before we take a look at the types of technology some financial organisations are using, remember that implementing technology isn’t ‘just because’. The real impact comes when you use technology to emphasise the following three areas:

• Customer insights: Produce correlations from dispensed internal data such as CRM, transactions, and investment stats.
• Intelligence: Merge customer insights with external data related to economic trends and behaviour.
• Customer engagement: Leverage data from customer insights and intelligence to deliver personalised customer experiences at the right time, through the right channel.

Here’s a teaser from our financial services whitepaper with three out of five ways organisations can use technology to drive innovation, build resilience, and be truly customer-focussed. Download the whitepaper to access the full guide.

[msce_cta layout=”image_center” align=”center” linktype=”blue” imageurl=”http://approjects.co.za/?big=en-us/industry/blog/wp-content/uploads/sites/22/2020/09/Lifestyle-image-of-adult-male-in-office-touching-screen-on-Surface-Book-3.-2.png” linkurl=”https://info.microsoft.com/UK-DTFIN-CNTNT-FY21-10Oct-14-Boostingtheinnovationofbankingbusinessmodels-AID-3024293-SRGCM3895_01Registration-ForminBody.html” linkscreenreadertext=”Download the report: Boosting the innovation of banking business models” linktext=”Boosting the innovation of banking business models” ][/msce_cta]

1. Public cloud infrastructure

If you’re looking to innovate into a digital-first model, you need strong cloud foundations. Public clouds can be deployed faster than on-premise infrastructures and there’s no extra cost of purchasing, managing, and maintaining on-premise infrastructures. Every employee can use the same application from any device, securely over the internet.

For finance organisations, compliance and security is very important. Microsoft Azure is built with a multi-layered security approach, with physical data centres, infrastructure and operations. There are 3,500 cyber security experts actively monitoring to protect your business assets and data. With over 90 compliance offerings, you can ensure valuable data is correctly safeguarded, and AI-driven security signals can also help modernise your security operations.

2. AI

Another way for organisations to empower employees is by using AI. AI-powered chatbots are ideal as a first point of contact, and can answer frequently asked questions. If customers need more help, they can move them onto customer service representatives.

AI can also be leveraged for knowledge mining and machine learning. This uncovers insights and analytics that can enable more informed business decisions. AI can test millions of ideas/scenarios per minute, uncovering insights and information such as credit risk scoring, identifying vulnerable customers, and interest rate changes. You can also test new business models rapidly. In the new normal, the ability to enrich existing data with external data will help build resilience within the organisation.

3. User experience

In the EY survey, 43 percent of respondents say the way they bank has changed. As customers are more used to digital ways of interacting, they expect financial institutions to adapt and innovate alongside them. Moving to mobile and web banking apps and leveraging new ways of delivering augmented experiences such as video conferencing, virtual reality, and augmented reality are key new business model enablers.

Citi traders, for example, is using the Microsoft HoloLens to see a virtual workstation that shows data as 3D images, making it easier to work and collaborate.

The new normal

In the new normal, competition will occur between business models rather than product and/or process innovations. Financial leaders should adjust their digital strategy to focus on supporting their customers with innovative ideas and empowering employees with new skills. When you think about how you’re planning on achieving your business goals in the new normal, make sure you build a skilling roadmap alongside. This will ensure there’s no skills gaps in your organisation.

Find out more

Download the report: Boosting the innovation of banking business models

Watch the on-demand webinar: How are traditional banks adapting their approach in a digital world?

About the authors

Tine Petric is a Specialist in the area of Applications and Infrastructure, advising organisations within the Financial Services Industry. He is passionate about the impact that technology can make in inclusive finance and ESG overall. Tine is also an avid tech blogger and guest lecturer at universities where he talks about Business Model Innovation and latest tech trends. Tine holds a Master Degree in Business Administration and Management from the HEC University of Lausanne, Switzerland

Christian Thier leads the Financial Services Account Team Organisation of Microsoft in Switzerland. He drives strategic and transformational partnerships with Banks and Financial Services firms across all segments, helping its clients to accelerate in digital business transformation. He has more than 20 years of working experience within the Banking, Insurance, Financial Services and IT industry in various roles. Before joining Microsoft, he was working at Interactive Data, serving as Managing Director and Board Member of Interactive Data in Switzerland, and Vice President Sales EMEA since 2005. Christian holds a Master Degree in Business Administration from the Goethe-University in Frankfurt/Main, Germany.

The post 3 ways the banking sector can innovate in the new normal appeared first on Microsoft Industry Blogs - United Kingdom.