It was 2 years into our ‘Gen AI journey’ before me and my long-time collaborator – Dan Aslet – began trying to transform Waterfield Tech with AI.
First, we had to figure out what this new kind of AI was capable of. We did a lot of experimenting and prototyping and built a lot of demos. We learned that the first 90% of a GenAI project is easy. Instant results. Almost no code. Real intelligence. Wow!
But getting from prototype to production is still a big lift. It’s the classic ‘Last Mile’ problem. The new stuff works great, but it’s the messy, legacy tech, processes, compliance issues, and economic realities where the real challenges lie.
Next we had to figure out what it meant for Contact Center and Customer Experience – the area we, and Waterfield Tech have been focused on for decades. That led to the concept of the AI-first contact center. It landed well and served us well. It’s the lens through which we view all contact center operations now. Check out the ebook we wrote on it here: The AI First Contact Center
At the same time, we had to transform ourselves to fully embrace AI as a co-worker. We could see the power of the models from our work on the AI-first contact center, but how could we integrate that power into our workflow, and into knowledge work in general? To do so we had to navigate the ‘Jagged Frontier’ of model capabilities. As Ethan Mollick puts it, based on his academic research with hundreds of Boston Consulting Group (BCG) consultants:
“AI has a jagged frontier. It is good at some things that seem very hard and bad at some things that seem really easy for humans.”
We began using AI all the time. At home, and at work. To research, write, code and create. Dan went through the most amazing transformation from tech aware director who lacked some detail when it came to systems architecture – and never wrote code – to a formidable technologist churning out AI agents, working applications, and beautiful user interfaces on a weekly cadence.
Only then could we start figuring out how to transform Waterfield Tech into an AI-first business. Because only then did we really understand what it would take to not just create AI capabilities, but integrate them into the business (solving the Last mile problem) and integrate them into individual workflows (solving the Personal empowerment problem).
In his latest substack article, Dan reminds us that before we get too carried away with grandiose plans to transform our organizations with AI, we should start by transforming ourselves and how we work: Why Personal Transformation Comes Before Business Transformation
In the article he shares 5 ways to get started. The first four are powerful ways to transform yourself: goal setting, knowledge acquisition, learning, and comms. Listen up, because its amazing what he’s achieved with this playbook!
The fifth is – for many – the ultimate goal: to get more done, but that relies on us being able to tell the AI how we do the things we do, so that it can do them for us.
For that, Dan reminds us that we need our own Standard Operating Procedures (SOPs) that define what we want to delegate to AI:
“clear, simple instructions that outline how tasks should be done, by who, and what to do if things go off track.”
Us humans tend to do a lot of things intuitively. Often the way we get stuff done is an unconscious competency – we’re consistently good at it, as a result of years of practice and experience, but we’d find it hard to break down each of the steps we take and the background knowledge and information we bring to the table.
Try it. Pick something you do that you’d like to delegate to AI and just write down how you do it. Don’t leave anything out.
If you’re like me, or Dan or anyone we’ve worked with on this, you’ll find it difficult. In your mind you cut corners and bring to bear deep experience without thinking twice. AI can’t do that. Not without the right context.
It’s the same with business processes. Your Standard Operating Procedures – if you have them – likely assume a bunch of background knowledge, experience and institutional context that an AI model doesn’t have. I bet there are multiple versions, too. Which is right? Is it up to date?
So just as you need to figure out how you really get stuff done before delegating to AI. You need to do the same with your business processes.
But wait, there’s a problem. This is a bunch of work. How are you going to find the time to do it? Is it gonna be worth it? And once you’ve got an SOP, how do you turn it into a form the AI will understand?
This is one of the central challenges in personal – and business – transformation with AI. It’s supposed to save time and money. It ought to up-level our knowledge work, and the products and services we deliver, but most of us are already struggling to hit targets and deadlines, and still leave enough time for family and sleep!
There are ways to overcome this organisationally, but like Dan says, you need to start with yourself.
How can you get from ‘AI curious’ to ‘It feels like cheating’.
I cover this, and a bunch of other things that will hopefully help you make progress on Dan’s playbook, in my new Substack – The Dualist.
My latest post: The Cheat sheet to Cheating with AI will help you get started playing with AI, making space for AI and finding the kind of focus that cuts through the firehose of information, updates, and opinions that overwhelm us all.
I’ve also got a post that goes deeper into Making space for AI where I outline 4 ‘AI action hacks’ that will help you find time where there is none, and a simple prompting framework: Simple prompting for Smart people.
And just in case you wondered, there’s a post on why we often say Please and Thank you to AI and why that’s actually a good prompting strategy!
Checkout The Dualist and hit subscribe while you’re there so future posts go straight to your inbox.
As a final reminder, if you haven’t already, grab your seat and next week’s LinkedIn live, where me, Dan and Fish will be talking about Model Context Protocol (MCP). As geeky as it sounds MCP should be on every CX, IT and AI leaders’ radar.
That same day, we’ll also be launching the updated version of this newsletter, which will be called: The Last Mile.
Why ‘The Last Mile’? Because like I said at the start of this email, the first 90% of AI-first transformation is easy. It’s the last 10% – integrating with your workflow, your business, and your contact center that gets us from Possible, to Practical, and ultimately Profitable CX, AI and IT solutions as Fish laid out in this article.
Kerry
PS: If you want a more regular dose of insights, follow or connect with me on LinkedIn for regular posts on conversational AI, mindset, and egg juggling, among other things!
PPS: You are building with GenAI right now, aren’t you? If not, what’s stopping you? Check out our blog on Gen-AI blockers, or sign up for a complimentary Strategy Workshop to help you get started.
If someone forwarded this to you, please subscribe yourself for weekly insights that’ll make you think differently about your IVR, voice, and chatbots.
Helping you get maximum ROI from conversational AI — whatever the platform.