Insight of the Week: Generative AI is not a Product

By Kerry Robinson

In their recent State of Data and Analytics report, Salesforce claim that 77% of business leaders are worried their company is missing out on generative AI's benefits.

The sense of concern among the contact center leaders I've spoken with recently is palpable. Everyone is under pressure to leverage generative AI.

So they're coming to us for advice. "We want to leverage the latest generative AI", they say. Then they ask:

• Which CCaaS platform should I buy?
• What knowledge management product do you recommend?
• Who's got the best agent-assist solution?

But these are the wrong questions. Unless you're OpenAI, Google, Microsoft, and the like, Generative AI is not a product.

It's a core technology, that powers products: Like a database. Or cloud computing. Or a programming language.

Right now, the products built with generative AI are shallow wrappers of core Generative AI techniques, like Retrieval Augmented Generation (RAG), which is the technical term for using AI to answer questions using content from documents, websites, and other sources. Or they're very specific applications of generative AI as specific functions in a product: like generating AI summaries from a call or chat.

But things are moving sooooo fast! If your CCaaS or Knowledge management, or CRM vendor has a specific generative AI feature that you like, great, use it.

But just because they have that specific feature, doesn't mean you're no longer 'missing out on generative AI's benefits' as 77% of business leaders believe they are.

The impact of generative AI can and should be significantly more than an auto-generated summary. Your aspirations should not, and need not be limited by the product roadmap of your contact center vendor.

Here's an example of the direction things are moving:

Back in July last year we launched Terry, my AI clone. Terry was trained on all of the emails in the Teaching Robots to Talk series. It's a simple but effective application of RAG.

I first built a concept demo of Terry in raw Python code. It took a few hours with coding assistance from ChatGPT because I'm an exceptionally mediocre programmer.

That showed what was possible: you could ask Terry a question, and he (or it?) would answer using relevant pieces of all these emails I write. It was impressive!

But it wasn't something we could deploy. We considered building our own solution but ultimately found a pre-built solution from a tiny startup. They had a ridiculously expensive cost per message, but we wanted to get Terry out there fast. So we went with it.

But now, OpenAI has its Assistants API and Microsoft has its co-pilots studio. I can - and have - rebuilt Terry on both platforms in about 15 minutes!

I can - and have - built conversation summarizers with the same tools. And tools to suggest responses to agents. And to delve deep into conversational data to extract insights.

The point is, you don't need to wait for your CCaaS, CRM, or Knowledge Management vendor to add these capabilities.? Nor do you need to buy a new CCaaS, CRM, or KM solution to get access to these capabilities. You can do it directly in the major AI platforms.

So what does this mean for your Generative AI aspirations?

My recommendation is: Don't go shopping. Go on a journey.

Go alone with your team, or with a vendor who can prepare you, travel with you, and guide you (and yes, we'd be delighted to do so!)

Understand what Generative AI is capable of: explore what's possible with Open AI assistants or Microsoft co-pilots.

Explore potential use cases in your contact center: how can generative AI automate interactions with customers, assist agents, and extract insights from your data?

And integrate what works with the AI exoskeleton approach that I covered a few weeks back.

In time, successful generative AI use cases will become established and may get integrated into the contact center technologies you buy and use. But that's a way off, and will always be a step or two behind the latest and greatest. If you want to get ahead of that 77%, take advantage of the 'composability' of the cloud to leverage core generative AI platforms.

Enjoy the exploration, and if you think you might benefit from an experienced guide, sign up for a complementary 90-minute strategy workshop?and we’ll help you plot a course that ensures your business is not missing out on generative AI's benefits.

Kerry Robinson is an Oxford physicist with a Master's in Artificial Intelligence. Kerry is a technologist, scientist, and lover of data with over 20 years of experience in conversational AI. He combines business, customer experience, and technical expertise to deliver IVR, voice, and chatbot strategy and keep Waterfield Tech buzzing.

Subscribe to Kerry's Weekly AI Insights