One of the mandates for the by us so coveted ISO 27001 certification is a retirement plan.

For me.

The goal of ISO 27001 is to ascertain proper company management. ISO spreads its wings across multiple areas of which retirement is but one. What the designers probably meant for us to do was to create policy regarding this subject, and to provide proof of enforcement. Easy peasy.

But it got me thinking.

We have a product which is a RAG index of text data together with an AI compiled FAQ. An AI powered chatbot as frontend. Clients create the index and the FAQ from any content:  their website, internal documents. Anything textual. We call it Korero, after the Maori word for conversation.

RAG stands for Retrieval-Augmented Generation. It helps AI focus on a particular dataset rather than on the whole of their training data.

What if we could create a RAG from me? No, I don’t mean through some hi-sci-fi neurolink. Just regular down-to-earth tech.

A company AI could answer any question and even come up with new ideas.

You see, as the closeted hoarder that I am, I’ve saved all my hard disks from all the computers I’ve ever owned.

All the digital products I ever created, all the proposals ever sent, everything, right there, in the vault I keep those drives.

Plus, crucially, the complete contents of my email boxes, the current, but also those long since abandoned. Every email I ever sent or received, along with their attachments.

It’s going to be hard, though.

Creating a comprehensive RAG index from this data would involve powering up the ancient hard drives, some needing connectors not been produced in decades, some badly corrupted. Identifying legacy file types and converting them, and moving them into a modern carrier.

A proof of concept conducted upon part of the data uncovered many, many issues.

Then size. My inbox is 70GB, and I had two terabyte disks in my old MacBook Pro. Regarding the sizes of the old disks, of course, the really old ones are tiny, but there’s an 8TB one too.

We’ll need a pretty big server.

But think of what we’d have.

Example. Right now, if we need a contract, we ask ChatGPT. We get a boilerplate contract back, of course we do. The AI has “read the internet“, levelled out all disparities, nothing special or personal about the results. For those unused to compiling contracts, undoubtedly a godsend. And much cheaper than hiring legal.

Imagine, though, if we had our own personal legal, legal that's deeply embedded in our company, knows all contracts we’ve ever negotiated. All the projects we ever did, all their pitfalls, all their costs, estimated and actual. That could create a contract intricately interwoven with our company’s specific skillset and values. Something special that clients would want to sign off on instantly.

Same for every area: project plans, documentation, slide decks, newsletters, company dashboards, marketing, write our emails, create our tickets.

We think a resource like this is going to be incredibly useful for us to build our dream: a highly automated Software Service Company as a Service.

But we could do more.

A personal AI could give instant access to all the photos, trinkets and other cherished things you just can’t throw out.

I say I have a hoarder’s mentality, but I think most have. One of our clients recently spoke of their ordeal moving to a new home without an attic. Where now to put all the stuff that they hadn’t looked at in years?

Me? Thousands of prints of our children from before the digital age. A suitcase packed with family photos. And, treasure above all treasures, a huge box of my dad’s 8mm home movies of us as children growing up in New Zealand. Prominently featuring my mother who died when I was eleven.

What if we could create a RAG index of it all?

We have a personal life that made us to who we are, formed us far more than than the professional life that we present so proudly.

Barring any neurolink miracle, my personal memories will be gone forever when I am. But we should be able to make a personal RAG index reality.

Have a question? Ask pseudo-me.