How Is Theta Using GenAI? Here Are Ten Use Cases

In light of this, we thought we’d share some of the cool stuff we’ve been working on at Theta, including ten use cases and learnings from those.

‍

First stop - how does our Innovation Lab use GenAI?

Our Innovation Lab has worked with GenAI since before ChatGPT launched, meaning we’ve had a good head start working with the technology – seeing what it’s really capable of and conducting various R&D projects.

‍

Use case #1

OmniSearch: We developed an internal knowledge tool that uses GenAI to index collections of documents and other data sources and allows users to ask questions about the data. 

‍

Use case #2

Test data generator:  An internal tool to generate fake synthetic test data for various scenarios – especially useful for testing analytics applications. Our Data and Insights team can create thousands of sample ‘customer records’ or any other record in minutes based on simple prompts. 

‍

Use case #3

Technology platform upgrader: Perform the initial mundane steps of migrating application code from one technology to another (or upgrade to a newer version). A combination of GenAI and sophisticated prompt engineering enables our developers to write new code reflecting the logic of the existing code. 

‍

What you can learn from these use cases:  

• Low-risk, internal use cases are a good starting point for GenAI projects.  
• Ask your busiest teams to list their unnecessarily lengthy manual tasks and prioritise how GenAI can help them.  
• And just as this article has been published, the team has created an AI code review step to our DevOps pipeline. New use cases are found all the time! 

‍

How does our Product team use GenAI?

‍

Use case #4

We infused GenAI capabilities into our self-service chatbot product, FAQ Bot.

We use it in two places.

First, for our new customers, to make initial set-up faster by reusing content they already have.

Secondly, to improve the ability of our chatbot to understand the questions people ask.  

How does it work?  

We use GenAI to help create the initial chatbot knowledgebase. We scan your website and generate a list of possible queries a website visitor might have (this is often in the hundreds) and answer them by summarising your existing content using AI. Pretty impressive and saves a lot of time!

To improve question understanding, we use AI to enhance the questions people ask with additional contextual information. This could be, for instance, helping the chatbot to understand how this question relates to an earlier question.  

‍

What you can learn from this use case:  

• If you’re infusing GenAI into one of your existing products (especially if it’s public or customer-facing), you must consider the risks.
• We know organisations are often concerned that letting a GenAI chatbot ‘loose’ on your website/social media channels can lead to mayhem and even lawsuits. So, an important consideration for us was ensuring the chatbot has the necessary controls in the background – mitigating the risk of an embarrassing or costly mistake caused by a rogue chatbot.
• In short, if you want to infuse GenAI into an existing product, you must do so with the necessary thought behind it. We’re happy to help with this development process.
• Figure out what the #1 challenge is with an existing customer-facing product, and then identify how GenAI can solve that particular challenge. Start small and scale.  

‍

Use case #5

We infused GenAI into our site sign-in and safety solution, EVA Check-in.  

If we refer back to the earlier point above, “start small and scale”, we identified that our customers needed a faster, more accurate way to manage suppliers – and something less manual.  

As a result, we now use GenAI to analyse documents uploaded by suppliers – like insurance certificates and other compliance documentation. We can detect if the insurance document is, in fact, an insurance document and if its expiration date is correct, as well as other custom checks as needed. 

What you can learn from this use case:  

• Where manual efforts are required, there's often an opportunity for a GenAI solution to be implemented.  

‍

How does our Dynamics 365 team use GenAI?

In addition to investigating Copilot applications, our Dynamics 365 team have been using GenAI to create custom solutions.  

‍

Use case #6

A sticky problem for Dynamics 365 Business Central users is uploading a spreadsheet journal into Business Central and manually mapping the fields accordingly. Our GenAI-infused extension Excel Importer allows you to a) upload this spreadsheet with a few clicks and b) handles this mapping automatically.  

‍

Use case #7

Each eDM sent by our marketing team was inundated with auto-replies, including those who had left their role, retired, were on parental leave, and otherwise.  

Using a custom-built AI model, we can ‘scan’ these auto-replies and automatically alert the contact owner of the change in status and deactivate the contact where necessary.  

As a result, our marketing team have less manual updating to do, we have a cleaner CRM, and the contact owners can remain up to date with the status changes of their contacts.  

‍

What you can learn from these use cases:

• Talking to end users is key to identifying specific sticking points, which can often be resolved using GenAI.
• Testing, testing, testing – a lot of work goes into ensuring a GenAI solution does what it promises. Without thorough testing, including end user testing, you risk it causing more issues than it solves.  

‍

Finally - how does our wider team use readily available tools like Copilot and ChatGPT?

Like many organisations, we worked quickly to establish staff guidelines for using ChatGPT and similar technologies and educated our team on the risks. We also updated our secure development guidelines to cover GenAI risks, and our staff use Microsoft Copilot enterprise (previously Bing Chat Enterprise) as their ‘general’ AI assistant as it’s protected from leaking corporate data. 

‍

Use case #8

Our developers make use of GitHub Copilot and GitHub Copilot Chat for GenAI coding assistance. It speeds up the coding process whilst still giving them control over the final code.  

‍

Use case #9

Selected staff have access to Teams Premium licences to enable them to use the AI meeting summaries. Initial feedback has been good. While not as fully featured as Copilot for Teams (for example, you need to wait for the transcript to be processed to get the notes), the AI notes feature does save time and this feature costs less than the full Copilot.  

‍

Use case #10

A pilot group of staff are using Copilot for Microsoft 365 to evaluate productivity gains and security implications. As a result, we developed our Microsoft 365 Copilot Readiness Assessment to ensure our customers can also activate the tool securely and effectively.    

‍

What you can learn from these use cases

• Preventing your team from using readily available AI tools such as ChatGPT will be virtually impossible. It’s fundamental to have the necessary measures to ensure they know the guidelines around using these.
• Using pilot groups of staff to test new GenAI technology, such as Copilot, is a good way to ensure a smooth transition.
• Shadow AI – the unsanctioned use of AI - is a growing problem. Our cyber security team can assist with any concerns you have about this.  

‍

Conclusion

‍

As you can see, the use cases for GenAI are broad and exist across the entire organisation. Many organisations already use GenAI more than they think, which is increasingly the case as more solutions and products release inbuilt GenAI capabilities.  

One thing’s for certain – as GenAI tech becomes more powerful, your use cases will accelerate.  

Organisations should focus on doing the necessary groundwork and developing a robust framework to identify, prioritise, and securely implement their GenAI-infused solutions. And if you need someone to partner with on that journey, we’re here for you.