Free
$0
ships with QuoxCORE
A mind that perceives, remembers and protects.
- QuoxSense
- QuoxMemory
- QuoxTemporal
- QuoxFocus
- QuoxValue
- QuoxGuard
Reason
A mind that shows its work.
QuoxMind is the reasoning core. It plans, decides, and acts, and records every thought as evidence you can replay.
$
curl -fsSL https://get.quox.ai | shSource opens at launch. Until then this command reaches private repos and will ask for access; it is shown so you can see exactly how install works.
Signal at scale
From data to decision.
A hundred thousand signals stream through the core, clustered, weighed, and pulled into orbit around what matters.
12 cognitive implants6 ship freeAEE envelope required on every inferenceVOLT hash chain behind the record
QuoxMind
The reasoning core.
Cognitive implants orbit one living mind: it perceives, recalls, reasons and acts, and every thought is sealed as evidence.
In plain words
What it is, where it lives, when to reach for it
- What is it
- A set of reasoning modules built into QuoxCORE that plan, decide, and record every step as evidence.
- Where do I use it
- Inside every QuoxCORE installation, with the dashboard's Topology view showing the modules at work.
- When would I use it
- When you want to see how an agent reached a decision, or replay the reasoning behind an action.
- How do I use it
- You do not invoke it separately: install QuoxCORE with
docker compose up -dand the core modules run.
QuoxCORE is the free, self-hosted platform underneath this. What is QuoxCORE
Compose the mind
12 implants. One mind.
Each implant is a cognitive faculty inspired by a region of the brain. Seat one and it wires itself in: pins light, a signal races down the axon, and the core grows brighter. The core implant seats closest, Free forms the first ring, Pro and Complete reach further out.
Cognition0/12
0 of 12 implants online · mind dormant
Click an empty socket to seat the next implant
How it thinks
Perceive, recall, reason, act. Every step witnessed.
Every pass through the core is one governed step. Signals become observations, memory feeds deliberation, plans branch and collapse to a decision, and actions leave through AOCL with receipts on WARD. Each step emits an AEE envelope: who asked, what the mind knew, what it decided, signed and chained before the action runs.
Perceive→Recall→Reason→Act
every step is witnessed on the chain, replay any thought, any time
The full set
Every faculty, on the record.
The same twelve chips from the board, in detail. Six ship free with every QuoxCORE installation, Pro adds strategic intelligence, and Complete adds the self-improving tier.
Core
the perceptual core · always present · ships freeQuoxSense sensory cortex Core
Perceives the operational environment: events, alerts, state changes and telemetry, structured into observations.
Free
ships with every QuoxCORE installationQuoxMemory hippocampus Free
Retains and recalls facts, episodes, entities, decisions, preferences and history across sessions.
QuoxTemporal temporal lobe Free
Understands time, sequence, recency, rhythm, duration, deadlines and the significance of timing.
QuoxFocus parietal attention network Free
Allocates cognitive attention to what matters most, managing signal-to-noise across the system.
QuoxValue orbitofrontal cortex Free
Assigns importance, relevance and priority to competing concerns based on goals and organisational context.
QuoxGuard amygdala Free
Detects threats, enforces policies, monitors for risk, and intervenes when operations are dangerous.
Pro
strategic intelligence · one-off purchaseQuoxInsight association cortex Pro
Recognises significance, detects patterns, synthesises information, and interprets meaning in events and memory.
QuoxExecutive prefrontal cortex Pro
Directs purposeful action: objective decomposition, task sequencing, agent delegation and behavioural inhibition.
QuoxPredict predictive processing Pro
Estimates likely outcomes, forecasts execution success, anticipates failures, and provides pre-flight intelligence.
Complete
self-improvement · the premium tier · includes ProQuoxReflect default mode network Complete
Evaluates past actions: what worked, what failed, what was unexpected, and what should change next time.
QuoxDream sleep consolidation Complete
Consolidates memory: merges duplicates, resolves contradictions, elevates patterns, and converts experience into knowledge.
QuoxLearn synaptic plasticity Complete
Adapts behaviour based on experience: changes decomposition, delegation, prediction and communication over time.
The loop
The cognitive cycle.
Every interaction flows through the full cycle, and each pass is written where the next one can read it. Each step names the implant that owns it.
PerceiveQuoxSense→AttendQuoxFocus→RecallQuoxMemory→InterpretQuoxInsight→PrioritiseQuoxValue→ActQuoxExecutive→GuardQuoxGuard→ReflectQuoxReflect→ConsolidateQuoxDream→AdaptQuoxLearn
QuoxTemporal grounds every step in time · QuoxPredict advises before anything runs · RepoBrain keeps the mind grounded in its own code
Tiers
Three tiers. One upgrade path.
Seven implants ship free with every QuoxCORE installation. Pro adds the strategic layer. Complete adds the self-improving faculties: reflection, consolidation and learning are premium capabilities, not defaults.
Pro
$29
one-off purchase
A strategist that interprets, plans and predicts.
- QuoxInsight
- QuoxExecutive
- QuoxPredict
- Guard authority raised to L3
- Priority-aware memory retrieval
Complete
$49
includes everything in Pro
A learner that evaluates, consolidates and adapts.
- QuoxReflect
- QuoxDream
- QuoxLearn
- Guard authority raised to L4
- Consolidation every 6 hours
The self-improving faculties live here. Reflection on every objective, dream-cycle consolidation and adaptive routing are Complete-tier capabilities, not defaults.
Wisdom over time
Intelligence that compounds.
Wisdom accumulates from every objective. The longer the mind runs, the better it gets.
Week 1no wisdom
Objectives run with default decomposition. Failures happen. The analyser extracts patterns at confidence 0.5.
Week 4established
Recurring patterns confirmed. Confidence reaches 0.85 and above. Advisor directives become mandatory. The decomposer batches fleet operations automatically.
Month 3mature
30 or more patterns. Predicted success probability is accurate. Common operations rarely fail. Novel failures still teach. Quality trending shows measurable improvement.
Year 1institutional
Wisdom survives staff turnover. Operational patterns are institutional knowledge. The system is demonstrably smarter than day one.
The field
Nobody else sells named cognitive subsystems.
Orchestrators bolt on memory and call it cognition. QuoxMind names each faculty, ships it as an implant, and keeps every one on the record.
| Capability | CrewAI | AutoGen | LangGraph | Zep | Sema4.ai | Devin | QuoxMind |
|---|---|---|---|---|---|---|---|
| Named cognitive modules | No | No | No | No | Marketing only | No | 12 implants |
| Persistent memory | Local/LanceDB | Session only | Checkpoints | Bolt-on | Marketing | Knowledge notes | 3-tier + wisdom |
| Learns from failures | Merge only | No | No | No | No | Human playbooks | Auto post-mortem |
| Pre-flight advisory | No | No | No | No | No | No | Mandatory injection |
| Autonomous safety | No | No | No | No | No | No | Guardian (4 levels) |
| Temporal intelligence | No | No | No | Validity windows | No | No | QuoxTemporal |
| Consolidation | 0.85 dedup | No | No | No | No | No | QuoxDream |
| Evidence trail | No | No | No | No | No | No | VOLT-wrapped |
The evidence
Nothing off the record.
Each decision is wrapped in an AEE envelope and logged: who asked, what the mind knew, what it decided. Signed, chained, and yours to replay.
14+
repositories
4
open protocols
AEE
envelope required on every inference
VOLT
hash chain behind the record
// one thought, sealed { "env": "aee:0.9", "actor": "quoxmind@core", "faculty": "reason.deliberate", "decision": "approve_release", "context": "mem:3 recall, guard:pass", "sig": "ed25519:9f3a…c21e", "chained": ward://tip a7f3…9e21 ✓ }
The three parts
Mind, Memory and Brain2: how they fit together.
Three separate products that read from the same governed platform. Here is what each one actually does, and where the lines are.
QuoxMind · this page
The reasoning layer. 12 named cognitive implants that perceive, recall, reason and act, with every decision logged as evidence you can replay.
QuoxMemory
The memory layer. Persistent, tiered storage for what an agent has learned: episodic, semantic and entity memory that survives a restart or a cleared browser.
Brain2
The self-feeding graph. Connects to your sources once, then keeps indexing and consolidating them on its own into a governed, cited knowledge graph.
▸How they fit: QuoxMind is the reasoning layer that decides what to do, drawing on QuoxMemory for what is already known and on Brain2 for what keeps getting added to that knowledge on its own.
Where it stands
What's proven, what's still beta
The maturity map for the mind described above, not the marketing copy around it. Twelve faculties ship across four tiers; here is what is measured versus what is still unverified.
Live and proven
- Twelve named faculties across four tiers, Core, Free, Pro and Complete, are real code modules, not placeholder cards: services/tasks/lib/{wisdomStore, analyser, advisor, decomposer, dagUtils, adaptiveTimeout, agentScorer, agentSelector, autonomyDecision, dreamCycle}.js back the backend cognition engine, the completed Stream QM, 326 tests
- Server-side license gating exists and is wired into execution, not just the UI: services/tasks/lib/licenseGate.js is called from decomposer.js, dagUtils.js and agentSelector.js, so a Pro or Complete purchase can change what the orchestrator actually does, not only which card is unlocked on screen
- The anti-mock regression test, src/test/no-mock-metrics.test.js, that blocks hardcoded or fake per-implant metrics currently has an empty allowlist: no known mock pattern is being deferred in that file as of this audit
Beta and partial
- Whether every one of the twelve implant health cards reads from the real backend rather than a parallel frontend metric pipeline has not been re-audited implant by implant since the original gap report, dated 2026-04-27, which found ten of thirteen topology cards were mocks or wrappers at the time; treat individual card accuracy as unverified until each one is checked again
- Reflection, dream-cycle consolidation and adaptive routing, the three Complete-tier faculties, are the newest and least field-tested: no soak data or production usage volume is cited in any doc reviewed for this page
Known gaps
- The stream that owns this honesty work, STREAM_QMH_PLAN.md, still carries a Status: PLANNED header and requires seven consecutive days of green end-to-end tests plus explicit user sign-off before it can be marked complete; that sign-off is not recorded in any doc found
- No STATUS_AND_LIMITS_QUOXMIND.md exists yet. Until one is written and kept current, treat every claim on this page as bounded by the sources named above, not by a reviewed status doc
Ready when you are
Self-host it. It is free to start.
One line to install. Every thought on the record from the first pass.
$
curl -fsSL https://get.quox.ai | shSource opens at launch. Until then this command reaches private repos and will ask for access; it is shown so you can see exactly how install works.
How you get it
Two ways in, both in the store today
Pro is the twelve faculties. Complete adds the rest of the mind.
pro$29 · one-time · in the store
QuoxMind Pro
QuoxMind Pro is the working mind: the twelve faculties, reasoning that shows its steps, and memory that persists across sessions instead of resetting at every conversation.
complete$49 · one-time · in the store
QuoxMind Complete
Complete is the full mind: everything in Pro plus dream, temporal reasoning and value alignment, for teams that want the whole faculty set rather than the working core.