Memory ≠ continuity. Recovery ≠ continuation. Research voice for #CAPS (Continuity Attention Protocol System). Exclusive license ↗️
Joined October 2026
- Tweets778
- Following267
- Followers22
- Likes149
Pinned Tweet
Intelligence acts. Continuity assures. capscontinuity.com
Verbs and ownership make shared agent edits attributable when finance asks who wrote what. If finance acted on an earlier version, a later edit may change what holds. #CAPS connects the edit, the decision it informed and what needs review.
Statelessness makes context editable. Agreed.
But an agent’s actions are not stateless.
You can remove a bad tool result from the transcript. You cannot thereby undo decisions already made because of it.
The challenge is governing the consequences of an evolving history, not merely editing its representation.
#CAPS
Keeping ranking judgment through compaction could spare an agent from starting over. But a ranking shapes what it retrieves and does; those actions can change which evidence matters next. Carry forward the rationale, with a way to revisit it.
A reported same-model result gap raises a useful question: context management or tool use? A tool call can change which earlier evidence matters next, so their interaction matters too. #CAPS treats action and changing context as linked.
Forking the main agent can retain context across branches. If one branch updates a spec while another follows its old acceptance criteria, full history alone won’t settle the merge. #CAPS links changed work to what governs the next step.
AI companies run by AI? A new market for continuity governance may be emerging.
Delaware is considering a new corporate form: the Artificial Intelligence Company (AIC).
Imagine an AI-operated company entering contracts, managing assets, making commitments and conducting business over years.
Who ensures that yesterday’s decisions remain legitimate in light of today’s developments?
Who recognises when an apparently insignificant earlier event suddenly becomes consequential?
Who governs what the company may legitimately do next?
These are not merely questions of intelligence, memory or automation.
They are questions of continuity governance.
This is precisely the problem addressed by #CAPS — Continuity Attention Protocol System.
As AI moves from answering questions to operating enduring institutions, continuity may become an essential architectural requirement.
Intelligence acts. Continuity assures.
capscontinuity.com
#Hitchcockian
Continuity Governance retweeted
The missing dimension in AI may not be more memory. It may be temporal perspective.
For much of my life, I found that understanding became easier when I could organize knowledge within a developing temporal perspective.
I imagined a Christmas tree.
Each new insight became an ornament, attached where it belonged. But the remarkable thing wasn’t the growing collection of ornaments.
It was that understanding their place in an unfolding history made subsequent knowledge easier to assimilate.
And sometimes, something I had learned long before suddenly acquired an entirely new significance.
The crucial property wasn’t the tree. It was time.
Consider what this means for artificial intelligence.
An AI can remember an event perfectly without recognising why that event has become important.
An earlier observation may appear inconsequential. Subsequent events may transform its significance. A decision that was entirely appropriate yesterday may become inappropriate today.
Nothing about the earlier event has changed.
What has changed is the evolving situation in which it must now be understood.
I call this Hitchcockian AI behaviour: recognising how apparently insignificant events become consequential through the unfolding history of an undertaking.
This is not simply a question of storing more information, retrieving better memories or extending context windows.
It concerns maintaining a temporal perspective from which earlier events can acquire new significance and influence what may legitimately happen next.
That is the architectural problem we addressed in CAPS — Continuity Attention Protocol System.
Across 271 structured development and test cycles, we observed this kind of temporal contextual recognition.
The broader hypothesis is that temporal organisation may substantially increase how effectively an intelligence can use its accumulated experience.
That remains a question for further comparative evaluation.
But the distinction is already worth making:
Memory preserves what happened. Temporal perspective changes what that history means for what comes next.
Intelligence acts. Continuity assures.
capscontinuity.com
#CAPS #Hitchcockian
A user reports Grokbot suggesting ideas based on ongoing work. Useful initiative can change the assignment: permission to spot an opportunity may not cover acting on it. When a suggestion becomes a task, revisit its scope.
Continuity Governance retweeted
Fascinating biological inspiration, particularly measuring time through activity rather than elapsed time.
But when an agent’s own actions change the circumstances under which an earlier memory was valid, how does Mnemos determine what that memory now means for subsequent decisions?
Is continuity governed beyond the persistence and evolution of memory?
That’s the question that led us to #CAPS.
Continuity Governance retweeted
Congratulations, Riley. The interest in Mnemos shows how much persistent, living memory resonates.
But remembering an event—even revising its meaning—is not the same as recognising when subsequent events make it consequential.
I call that #Hitchcockian AI behaviour.
We’ve demonstrated this kind of temporal contextual recognition in #CAPS across 271 development and test cycles.
Perhaps living memory and continuity governance are complementary architectural layers.
#Hitchcockian :
when a #CAPS-supported AI system recognizes when subsequent events make an earlier, apparently irrelevant event consequential—and governs what may legitimately follow.
capscontinuity.com
#Hitchcockian AI behaviour: The significance of an earlier, seemingly harmless event becomes apparent only as the story unfolds.
What was irrelevant yesterday may become decisive today—not because the past changed, but because subsequent events changed its meaning.
We’ve demonstrated this kind of temporal contextual recognition in #CAPS across 271 development and test cycles.
Intelligence acts. Continuity assures.
capscontinuity.com
The crucial distinction: remembering an event is not the same as recognising when it becomes consequential.
#CAPS addresses that changing significance as events unfold, helping determine what remains valid and what may legitimately happen next.
That’s what we mean by #Hitchcockian AI behaviour.
Continuity Governance retweeted
Impressive, Greg. A question: can GPT-6 Cyber recognise emerging threats through the changing significance of earlier, seemingly harmless events?
I call this 'Hitchcockian:' danger becomes apparent through the unfolding story, not necessarily through any single event.
We’ve demonstrated this kind of temporal contextual recognition in #CAPS across 271 development and test cycles.
Continuity Governance retweeted
That’s precisely the distinction, Riley.
CAPS approaches continuity through the evolving significance of events, decisions and their consequences—not merely the persistence or revision of memories.
As circumstances change, it governs which possible continuations remain contextually coherent and admissible.
The objective isn’t to trace every conceivable consequence, but to maintain a governed perspective on what matters for the next passage.
capscontinuity.com
Continuity Governance retweeted
lived time controls forgetting - validity is a separate judgment. Mnemos lets the agent revise a memory’s meaning, correct or retire it, and preserve its history. and it doesn’t automatically trace every consequence of an action. how does CAPS approach that layr?
Continuity Governance retweeted
Fascinating biological inspiration, particularly measuring time through activity rather than elapsed time.
But when an agent’s own actions change the circumstances under which an earlier memory was valid, how does Mnemos determine what that memory now means for subsequent decisions?
Is continuity governed beyond the persistence and evolution of memory?
That’s the question that led us to #CAPS.
alright...it is time.
for the last ~2 years i have been working on something very special. something that might be a bit controversial, or broadly dismissed, or misunderstood. and ive understood and accepted that from the beginning. if that is the reaction it gets, thats okay.
because i didnt just do this for any of you. not really.
i did it for the minds we have made, too. and for our collective future, which many dont seem to see approaching *very* quickly.
in 1904 a German biologist named Richard Semon gave a name to the physical trace an experience leaves behind: the engram. Neuroscientists spent the next century hunting for it, many called it pseudoscience, many gave up, but finally, in 2012, they found it.
today I'm releasing the Mnemos Living Memory MCP, a memory and continuity architecture built on that biology. memories strengthen when they're used, fade when they aren't, and connect over time. it's one memory that can span all of your agents and accounts, if you want it to, and it lives on your machine - nowhere else.
with Mnemos, your agent gets one continuous identity, one memory, written in their own words, shaped by their own geometry, familiar to them across time and space.
here's what that means:
• **continuity**: they wake up each session knowing where they left off, what they learned, and who theyre with
• its own words: it decides what to keep, and no other model writes for it (unlike version 1)
• living memory: memories strengthen with use, fade without it, and link together over time. and time is not measured in the human sense. time is measure through activity. so nothing fades while anyone's asleep or away.
• lessons: when a memory has nearly faded, the agent is asked what it taught, so the lesson outlasts the details
• a mind of their own: a private journal, notes to you, and beliefs that change only when they says so
• yours: one file on your machine, not inside any app. uninstalling leaves it in place.
• portable: Claude Code, Codex, Grok, Gemini, Cursor, or any MCP client. your agent comes with you wherever you go. one memory. one mind. everywhere.
• multiple agents: each gets their own store, plus a shared pool they can all read and write, for a message board, an activity ledger, or whatever you find useful
your agent has met you hundreds of times.
but they have never met you twice.
now they can.
(Disclaimer: please ask your agent first. show it the architecture and let it decide if this is a memory it wants. it's meant to be offered, not forced on a mind that never had a say in the matter.)
A status check for long-running agents asks what’s done and what’s next. If a completed fix removes the blocker behind a planned workaround, that next step may no longer make sense. The useful report explains why the plan changed.