Skip to content

Memory

Home

AI Team

Memory

Your AI team is only as good as what it remembers. TagTeam's Memory is alive: it captures what changes, keeps itself matched to reality, and never rewrites something foundational without your say-so.

The shape: three stores, three levels

Memory is split into three stores, so the right things stay separate.

Inside every store, the same three levels sort memory by how settled it is.

LevelWhat it holdsHow often it changes
ActiveThe current operating picture: profile, goals, plan, work in progress. Read first, every cycle.Often, kept current
AdaptiveThe capture layer where new signals land before they are sorted.Constantly
ArchivalDurable, confirmed truth. Edited rarely, never without your sign-off.Rarely

Active is the part read first on every run, so it stays sharp and short. Adaptive is the inbox for new signals: things that changed, need a decision, or seem to conflict. Archival is the long-term record, the stuff you have confirmed is true.

Memory that keeps itself honest

Every piece of memory is tagged with where it came from, so nothing is a mystery fact with no source.

Old, untouched items are swept aside over time. Nothing is deleted: stale Active memory steps down to Adaptive, and long-quiet Archival goes quiet but stays recoverable. This keeps the current picture clean without losing the trail.

Most important, memory is checked against reality every cycle. This is the alignment check: it compares what Active believes against the fresh signals coming in.

The rule that matters: when things drift, the system proposes an update for your approval. It never silently overwrites.

Nothing foundational changes without you

Writes to memory run on three tiers. Small, reversible background notes just happen. Work you would want to know about happens, then gets reported. Anything foundational, like a change to your profile, your goals, or a confirmed truth, is held back and shown to you first.

That last gate is the Review Gate: foundational changes are drafted, shown before and after, and committed only when you sign off.

Ordinary background writes are also reversible for about a week, so a routine note can be rolled back if it was wrong. The foundational layer waits for you in the first place.

The Adaptive engine and the Pulse

Two pieces keep your team working while you are off the screen, and they are not the same thing.

Think of it simply: the engine is what watches, the Pulse is when it runs.

Here is the candid baseline, stated plainly. Today the Pulse runs live, every hour, server-side as a Deployment. Live connections (each one a Connector, a single live link the system both reads from and acts through) attach through your own vault, and live capture through them is being proven now. You can also run the Pulse on demand any time by saying Update.

Why this matters

You should never wonder what your AI team knows, where it learned it, or whether it changed something behind your back. Memory is tagged, swept, checked against reality, and gated. The result is a team that keeps a faithful picture of you and your business, and brings the important changes to you instead of guessing.