A vision document for a living system that turns raw life and work signal into memory, better questions, clearer decisions, protected focus, and executed action.
You’re building a living operating layer between your life, your work, your memory, your ambitions, and the systems you want to bring into the world. Hermes is not meant to be a chatbot you ask questions to.
Or more emotionally: a second nervous system for a founder trying to become the person capable of building the future he sees.
Each piece maps to a body-system metaphor: active operator, memory graph, ambient capture, heartbeat, nervous system, and attention steering.
The active layer: listens, remembers, prompts, delegates, runs tools, schedules crons, checks systems, asks better questions, and turns vague intention into concrete workflows.
chatbot → chief of staff → operating partnerA durable graph of people, projects, ideas, commitments, open loops, decisions, events, emotional salience, recurring patterns, and original thinking.
make context compoundThe black box recorder. The deeper idea is not transcription: life should become queryable without requiring perfect deliberate capture.
raw transcript as evidenceMakes the system alive: morning focus, system health, market awareness, audio reflection, GBrain maintenance, open-tabs review, reminders, watchdogs, and nudges.
AI re-enters the roomTelegram, SMS, Sendblue, audio capture, and future WhatsApp/Gmail/iMessage/Keep. Context flows through the surfaces where life already happens.
surface → system → actionNot a task manager: a consciousness manager. It asks what deserves Connor’s limited conscious attention next.
attention steeringopen app → enter data → manage list → remember to check dashboard
live life → context is captured → meaning is extracted → decisions surface → actions are taken → memory compounds
Built around lived context, not empty prompts.
Durable enough to remember people, projects, commitments, and patterns.
Able to act through tools, schedules, scripts, and delegated agents.
Present at the right moments without becoming surveillance or noise.
Raw private context stays protected; canonical summaries and decisions compound.
Prompts better questions and converts messy human life into structured momentum.
Category thesis: from productivity software to personal operating infrastructure; from apps that store information to agents that metabolize context.
The personal Hermes/GBrain stack is the dogfood version of the company thesis. Connor is using his own life as the first messy integration environment: voice, messages, notes, email, finance, tabs, crons, reminders, code agents, market awareness, open loops, and the founder identity transition.
Whoever owns the context layer can help steer the work layer.
A lot of human potential is lost in the gap between what someone notices, says, intends, remembers, decides, actually does, and what their systems reinforce. Hermes is becoming a bridge across that gap.
The system should help you become more capable — not just complete tasks, but hold more complexity.
Reduce latency between signal and action: idea → note → candidate → decision → task/cron/agent → infrastructure.
Make reality visible: what is open, blocked, costly, repeating, meaningful, avoided, and deserving attention.
The system is not just assisting current Connor. It is scaffolding future Connor: less hidden drag, faster decisions, cleaner context, protected deep work, more delegation, earlier pattern detection, chaos converted into systems, and AI used as leverage rather than entertainment.
What would the founding engineer technical leader do here?
What are you holding onto that no longer belongs in the identity you’re becoming?
Voice, messages, notes, email, finance, links, tabs, code, meetings.
Evidence preserved without indiscriminate publication.
Signal becomes candidates, tasks, facts, decisions, reminders.
Entities, salience, open loops, deadlines, original thinking.
The system returns at the right moment.
Protected attention, executed work, code agents, closed loops.
Life and company operating patterns improve.
Dogfood becomes Cloop thesis and product direction.
The danger is building the operating layer instead of using it to move Cloop and life forward. The system must cash out in one of five things: clearer decision, protected focus, reduced drag, executed action, or reusable product insight.
In the future, serious people and teams will not manage their work through scattered apps, lost conversations, stale dashboards, and manual todo lists. They will have context-aware AI operating layers that continuously ingest their world, preserve memory, surface the right questions, coordinate action, and help them become more capable over time. These systems will be private where necessary, agentic where useful, and deeply integrated with the communication surfaces where life and work already happen.