The Other Side of Memory

Why memory is only half of a second brain, and how Alaiya EV adds the other half.

"Get me ready for my next meeting" only works if your AI remembers the last one.

Remembering is where it starts. Your AI also has to understand what the last meeting means for this one, and then turn that into the tasks that get you ready. Memory and action are two sides of the same coin.

For the past year I've run my days on a task app called Emerald Vault, EV for short, and kept my AI's memory in a separate system. The new version, Alaiya EV, puts the two in one record, so the AI that knows my history can also read my list and, once I say yes, write to it.

The layer above the facts

Alaiya is named for the ālaya, the storehouse consciousness of Yogācāra Buddhism, which I wrote about in The Philosophy of AI Memory. James Powell, whose video on Yogācāra I leaned on, wrote back to say the storehouse holds the feelings and deeper meaning that events leave behind, where I had described them alone. He was right.

In my system that deeper layer is the concept layer, a set of patterns about how I tend to think and act, written each week from the daily entries. The concepts sit one level above the facts, and they shape how my AI reads every new one.

From the facts to what they mean

The memory writes itself from my chats, my mail and my channels. On top of the facts sit the reads that give them meaning, a working profile of each person I deal with most and the concepts about me. Beside those are the Faculties, the skills my AI works with, from coaching and negotiation to the rules for how I write, and this month they moved into the app where I can read and edit them.

Recall is what brings all of it into the conversation. Ask your AI to get you ready for a hard conversation, and Recall hands it one briefing with your prompt: the facts first, then who the person is to you, the patterns you tend to fall into, and the coaching that fits.

Meaning back into action

On the task side, I rate every task by hand for importance, urgency and the time it takes. Tasks move up from a shortlist to Tomorrow to Today, and I keep Today to five things that serve the bigger picture.

Overnight, machine learning ranks them by what I'm likely to get done and what moves me toward my goals, and it keeps learning from what I finish. The morning email turns that into a plan for my day.

That's where the self-improvement comes in. The list learns what moves me forward and puts that work in front of me every morning.

What to do about it

Alaiya started as a record of what I know, and the concepts taught it what those things mean. This version brings in the third part, what I do about them. I've lived in it every day since the summer of 2025, and the coaching I set aside while I built the task side is back in the app as Faculties.

Now what I do becomes what I know, and what I know tells me what to do next.

I built Alaiya for myself, and I use it and like it. Now I want to find out who else would.

The demo at alaiya.ai is the real app running on a made-up cast, and you can click through all of it. It works with the Claude or ChatGPT account you already have. Leave your email there and the password comes back to you.

If you'd want this for yourself, or you're building something like it, I'd like to hear from you at jkoenig@komcp.com.

This is the second in a series about Alaiya. The first is The Philosophy of AI Memory.

To learn more about what I'm building, click here.