Salesforce chief executive Marc Benioff spoke with the tech blog platform Stratechery last week. The interview reveals what AI agents really mean for business and at the same time, it’s a masterclass in positioning vs competitors. It also illuminates a key factor in deciding which contact center AI platform is right for you: suite, custom, or native?
The impact of AI agents
Benioff covers the importance of AI agents in transforming business. If you’re not thinking about an agentic AI layer for your business – he says – you’re behind. We’ve been talking about the AI-first contact center and AI agents for 6-months now, so you’re ahead of the game here!
He talks through example use-cases in healthcare: streamlining patient care with AI agents that act as advocates, coaches, guides and assistants to help patients access and navigate complex healthcare systems and treatment regimens.
He explains how AI agents can support customer service and technical employees in telecoms, providing immediate access to relevant data, troubleshooting steps, and solutions.
And in retail, he points to how textbook seller Wiley handles peak demand during back-to-school season with the efficiency, and real-time scalability AI agents offer.
If you want more examples, check out the replay of my recent AI-first contact center webinar.
Marketing master class
Benioff talks about how early mentors pushed him to add sales and marketing to his prodigious repertoire of tech skills.
He reminds us that you need to either be the leader or position yourself against them. Salesforce is the second largest software company after Microsoft, so it’s no surprise that he goes on to transparently position Salesforce and their new Agentforce platform against Microsoft.
He calls out Microsoft’s copilot for being rubbish. It is. He deftly extrapolates from the underwhelming experience of copilot to the rest of their AI being just as bad, and Salesforce being better. But there I think he’s wrong. The agentic AI tech is the same across Salesforce, Microsoft Azure AI services, and even copilot. Nobody has a monopoly on AI science, technology, or methodology. It’s how you use it, that matters.
AI-first business
The essence of an AI-first approach to contact centers – and business in general – is to leverage AI agents to get stuff done: answer calls, train agents, evaluate experiences, enrich data, analyze it, etc.
To do this, you need to create AI agents.
An AI Agent has three main elements: Instructions, an AI model, and tools:
1) Instructions: give your agent a role, an objective, tasks to complete, and instructions on how to do that.
2) An AI model: lets your AI Agent figure out how to perform its role, to deliver on its objective, by performing tasks, according to the instructions.
3) Tools: give the AI Agent to access systems and data to get stuff done.
Microsoft Azure lets you build Agents: you can build “Assistants” (the Open AI version) or “copilots” (The Microsoft version) by giving them instructions, allocating an AI model, and providing access to tools.
It’s extremely flexible. You can build anything. But it’s quite manual. You’ve got to do all the plumbing. This is how we build most of our AI agents that we call xGPT solutions. This is the custom approach.
Salesforce’s Agentforce lets you build Agents too. The details are less clear at the moment, but no doubt they combine Instructions, and AI model, and tools. It will be much less manual. A lot more will be done for you, so most things will be easier, but some things will be a lot harder. You get speed to market and ease of deployment, but you pay the price of not being able to use any model, any framework, or any tools quite as easily as with Microsoft. This is the suite approach.
And then we’ve got Co-pilot. This is the AI Agent built into your Microsoft 365 subscription. It sounds like what Microsoft Azure and Agentforce are offering, but really, it’s not. It’s the lowest common denominator solution. It’s an agent that’s supposed to help everyone with everything they might want to do with Microsoft desktop products and it fails at most of them. This is the native approach.
The real message here is that there are different ways to build an AI-first business and an AI-first contact center. It’s all about trade-offs. The main trade-off being between Custom and Suite. Beware of the promises of ‘native’ solutions. Just as Co-pilot is a very poor comparison with what’s possible on Azure and Agentforce, the ‘AI capability’ built into your CCaaS is likely a very poor alternative to what’s possible with an AI-focused custom solution, or suite. Use it for the basic stuff, but don’t expect to realize our vision of an AI-first contact center or Marc Benioff’s vision of an Agent first business with a lowest common denominator solution.
Kerry
PS: You are building with genAI right now, aren’t you? If not, what’s stopping you? Check out our latest blog on gen-AI blockers or sign up for a complimentary Strategy Workshop to help you get started.
PPS: If you want a more regular dose of insights, subscribe to our weekly email that’ll make you think differently about your IVR, voice, and chatbots.
About the Author
Kerry Robinson
VP of AI Strategy
Waterfield Tech
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.
