Column A year and a half into the explosion of AI fueled by ChatGPT, the hype and fear of missing out has begun to thin just enough to make out the shape of two starkly different visions for AI: one that imagines using it to replace people and the other that wants AI to enhance people.

I spend a lot of my time working with the latter - and don't have much time for folks who see AI only as a means to write human effort out of the economic picture. Everything I've learned about AI tells me that as you add automations, you increase the need for oversight.

Hallucinations are only a small part of that. The vagaries of everyday life also complicate matters by always throwing up edge cases that no automation - however clever - can anticipate or resolve without human assistance.

That won't keep folks from seeking the fools' gold of the "one-man band": a massive entity, controlled from a central point, by a single person, perhaps informed by AI. Ironically, leaders of such orgs often prove their fitness for the job by being efficient cutters of costs and people, rather than capable of devising and executing growth strategies.

In the real world of SMEs, the drive to automate is hoped to massively increase the productivity of staffers and thereby grow the business. That sounds easy, until you discover that workers fill unique roles within smaller organizations - and many don't know what they do that truly delivers value.

That gap comes because jobs involve explicit work as well as many things we don't even know we're doing. That tacit work keeps organizations alive - if it gets lost in a drive to automation, the organization will suddenly founder.

It's hard to identify and measure tacit work. Nearly a century and a half ago, Frederick Winslow Taylor's principles of "scientific management" developed into practices that recorded, systematized and scaled mass production processes. Scientific management works well with processes generating physical outputs, but struggles to codify processes dependent on communication and collaboration. On the shop floor, Taylorism; in the office, tacit work remains supreme.

To capture the workflow within an office, you need to be down in it - keenly observing both the processes and all the interactions that shape those processes. SMEs seeking to amplify employee productivity must start here with a little light anthropology - or even a full organization analysis - that explores their own habits.

A proper analysis, of course, costs more than most SMEs can afford to pay.

But what if there were some way to automate the observations? Could that make it more accessible to SMEs who need to do that first bit of work, so they can understand how to increase worker productivity?

That conversation has been consuming my business partner and me over the last months, as we've learned what it takes to bring the benefits of AI to smaller firms.

Then, last month, Microsoft dropped a tool that no one had expected or asked for - and which seemingly no one needs.

That tool is Recall, the software that records everything done with a PC and regurgitates it.

Much has been written about the manifold privacy and security issues Recall poses. To me, so much seemed weird about Recall that I never really saw the sense of it.

My clear-eyed business partner saw through it immediately: "It's Robotic Process Automation."

The most straightforward path toward automating any process is brute force: capture it in detail, then replicate what you've captured. Recall provides the perfect tool to capture all interactions with the computer, generating a compact plaintext database, organized by application and time - the sort of thing that's perfect for ingestion by a large language model, like Copilot.

Imagine using that Recall database with a prompt like "This is a sequential series of user interactions. Explain in a detailed, step-by-step analysis, what the user is doing, from beginning to end."

A few minutes later, any SME might find themselves reading an account of their key processes, neatly laid out and ready for automation.

It's the one blindingly obvious use case for Recall - even if Microsoft can't say so publicly. Why? Because most of Redmond's enterprise clients are obsessed with the fantasy of a "one-man band". Let slip the real purpose of Recall, and every worker in a mega-corp will suddenly realize that - with every keypress and mouse click - they're automating their way out of their jobs.

For SMEs, it's a very different story. Recall offers an inexpensive path toward process capture in the office, reducing the cost of bringing AI into the business. Recall opens the door to a range of agent-based workflow automations for SMEs - that Microsoft will happily host on its own cloud.

How long until Microsoft admits the real purpose of Recall is to give businesses the tools they need to improve staff productivity by automating their everyday chores? SMEs will hear that message and embrace Recall - even with all of its privacy and security concerns - to race past businesses so centered on cutting costs they would rather bleed out than grow. ®

https://www.theregister.com//2024/06/12/microsoft_recall_sme_benefits/