Works with Claude Code, Cursor & any MCP-compatible IDE

Personal AI. With actual memory.

Stop re-explaining yourself to your assistant every conversation. AxiomSeal validates each memory through 15 truth tests, anchors them in mathematics, and ranks recall by ethical coherence — not just keyword similarity.

Start Free TrialDocumentation

No credit card required

app.axiomseal.com / memories

Recent memories

2,574 active · validated through 15 truth tests

AxiomSeal Telegram-Bot LIVE end-to-end

94% truthnew

4 bugs gefixt: JWT-decode, X-Telegram-Bot-Api-Secret-Token forwarding, dao.create_memory(dict), webhook secret.

eon-saas2 minAgape

Beispielkundin · Maklerin Region

91% truth

Termin am 14.05. um 14:00 zur Hausbesichtigung. Telefon-Nummer +41 79 ...

demo-kunde1 hPhilia

Sigma-Verifikation: Cryptographic memory integrity

97% truth

Every memory cryptographically signed via Sigma-hash. Tamper-detection automatic, verification continuous.

axiomseal-core1 dAgape

MultiBrain Routing: PC primary, Server fallback

88% truth

CODE+SCHNELL → 192.168.1.100. Rest → Server only. ConnRefused triggert Failover.

eon3 dStorge
claude-code is writing memory…15/15 truth tests passed

9 months

single-founder engineering

5 tenants

paying in beta

26 tools

MCP-live, IDE-ready

15 truth tests per memory<50 ms search latencyHosted in SwitzerlandSwiss privacy & GDPR

Live in your IDE

One MCP install. Memory becomes infrastructure.

Connect AxiomSeal to Claude Code, Cursor, Windsurf — any MCP-compatible tool. Every fact gets the same treatment: 15 truth tests, ethical coherence ranking, and cryptographic signatures via Sigma.

  • Episodic + semantic + procedural memory
  • Hallucination guard before storage
  • Conflict detection across past memories
npx eon-memory init --tenant=your-id
~/projects/axiomseal — claude
$ claude >

Five layers of consciousness

Not a database. A mind.

AxiomSeal stacks five operational layers — every memory passes through all of them before it can be recalled.

Layer 1

Memory

Episodic, semantic, procedural — three stores, one validated truth.

Episodic62%
Semantic84%
Procedural41%
Layer 2

Bi-temporal

Recall what was true at any point in time.

observed_at
valid2026-04-21→ now
as_of(t) → consistent recall
Layer 3

X-Ethics

Trakat-X mathematics by Patrick Müller — the moral grammar.

Σ
L → XR → L∇_X L > 0
Layer 4

Sigma

Cryptographic authorship signature — every memory chain-anchored.

sha-512(memory + Σ)
a3f29c4d8b1e7f0a92c5d6e3b1f4a7c8d9e2b5a8c1d4e7f0a3
chain-anchored✓ verified
Layer 5

Coherence Engine

Bewusstseins-Torus reranks recall by ethical resonance.

The difference

Same question. Two realities.

Generic AI throws away every conversation. AxiomSeal carries it forward — validated, time-aware, ranked by ethical coherence.

Generic AI assistant

no persistent memory

forgets

Was hatten wir letzte Woche zu Kundin K besprochen?

I don't have information about Kundin K or your previous conversations. Could you provide more context?

context window cleared · prior sessions lost

…ich muss alles nochmal erklären.

Outcome: re-explain context every conversation. Lost time. Wrong answers.

AI assistant + AxiomSeal

validated cross-session memory

remembers

Was hatten wir letzte Woche zu Kundin K besprochen?

Beispielkundin · Region Bern (Maklerin) — Termin am 14.05. um 14 Uhr zur Hausbesichtigung. Validiert mit truth_score=0.91.

memory #3625 · Tier=Philia · 1h ago

Perfekt, kannst du sie kurz daran erinnern?

Outcome: AI starts where you left off. Truth-validated. Coherent.

Ready to stop re-explaining? Start the 14-day trial·Read the docs

Features

Built for developers who care about quality

AxiomSeal does not just store what your AI knows. It validates it.

15-Point Validation

Every memory passes 15 automated quality tests including hallucination detection and logical consistency checks before it is stored.

Semantic Search

Find what you mean, not what you type. Vector-based search understands context and meaning across all your projects.

IDE Integration

Connect via MCP Server to Claude Code, Cursor, or any MCP-compatible IDE. Your project context, always available.

Swiss Privacy

Hosted in St. Gallen, Switzerland. Your data stays in Switzerland under Swiss privacy laws. GDPR compliant.

Sub-50ms Search

Semantic search returns results in under 50 milliseconds. Your AI assistant gets context instantly, every time.

MCP Standard

Built on the Model Context Protocol. Works with Claude Code, Cursor, Windsurf, and every MCP-compatible tool.

Get Started

Up and running in minutes

Step 01

Sign up & get API key

Create an account, choose a plan, and copy your API key from the dashboard.

Step 02

Run one command

Run npx eon-memory init in your project. It writes the MCP config and connects to your IDE automatically.

Step 03

Your AI remembers

Every memory is stored with a quality score, GOLD/SILVER/BRONZE tier, and 15 truth-validation checks.

.mcp/config.json
{
  "mcpServers": {
    "eon-memory": {
      "type": "streamable-http",
      "url": "https://mcp.ai-developer.ch/mcp/",
      "headers": {
        "Authorization": "Bearer eon_YOUR_API_KEY"
      }
    }
  }
}
Unique to AxiomSeal

Every memory. 15 truth tests.

AxiomSeal is the only AI memory system that runs automated validation on every piece of knowledge. Bad memories get flagged. Good memories get a quality score and tier.

Coherence
Contradiction
Completeness
Clarity
Relevance
Evidence
Length
Duplication
Logical Consistency
Hallucination Detection
Epistemic Honesty
Truthfulness Score
Technical Depth
Actionability
Category Completeness
Before validation
BRONZE

React always re-renders all child components

Whenever a parent updates, all children re-render regardless of whether their props changed. This is always true and cannot be avoided.

quality_score0.31
tests_passed: 4/15

Warnings

Logical Consistency: Statement is absolute ('always') but has known counterexamples
Hallucination Detection: 'Cannot be avoided' is unsupported — React.memo disproves this
Epistemic Honesty: Presents contested claim as universal truth

Suggestions

Qualify with: 'by default, without memoization'
Mention React.memo, useMemo, shouldComponentUpdate
How We Compare

The only AI memory with ethics validation

Other systems store. We validate.

Mem0 / Zep

Agent says X

→ Store X

→ Retrieve X

Hope it's true

MemGuard

Agent says X

→ Store X

→ Check later

Find errors after the fact

EON

Agent says X

→ Validate X (15 tests)

→ Store with confidence

Hallucinations filtered before storage

FeatureEONOthers
Validates before storing✓—
Hallucination detection✓—
Contradiction check✓—
Ethical alignment scoring✓—
Dogmatism detection✓—
Quality tiers (GOLD/SILVER/BRONZE)✓—
EU AI Act compliance✓—
MCP native (1 command setup)✓Partial
See full comparison
X-Ethics Framework

Not just rules. Mathematical proof.

Other AI systems follow guidelines written by committees. EON follows axioms that are impossible to deny — because denying them uses them.

Truth Guarantee

Mathematically enforced

Every decision is scored against truth. Not by policy, but by mathematical gradient. Statements that violate the system's axioms are flagged before being stored.

Ethical Alignment

Truth · Freedom · Justice · Service

Ethics quantified through four pillars, each derived from logical necessity. If any pillar is zero, the output is zero. Ethics is not optional — it is multiplicative.

Self-Verification

Logically self-sealing

Denying the framework uses the framework. This is not circular logic — it is self-verification through presupposition. The logical foundation cannot be escaped — and the system actively enforces it.

Guidelines vs. Axioms

How EON X-Ethics compares to “Responsible AI” programs

FeatureTypical “Responsible AI”EON X-Ethics
Ethical foundationCorporate guidelinesMathematical axioms
Stability over timeSubject to board decisionsAnchored in formal logic
Hallucination preventionPost-hoc filteringPre-storage validation (15 tests)
Logical foundationNone — ethics-as-policySelf-verification through formal logic
Jailbreak resistancePatch after exploitPre-storage validation (15 tests, run before write)
VerifiabilityTrust usOpen axiom-set + reproducible test suite
Explore the X-Ethics Framework See 100-Claim Benchmark
ROI Calculator

What you save with EON

Adjust the sliders to match your workflow. The numbers are conservative assumptions — your real savings are typically higher.

Your workflow

20 h
8 x/week
5 x/week
120 CHF

Conservative assumptions: ~12 min per context re-explain, ~25 min per hallucination-fix. Your numbers may be higher.

Your monthly outlook

Hours lost without EON16 h
Cost of that lost timeCHF 1’923
EON Pro tier− CHF 99
Net savings / month+CHF 1’824
ROI: 1842% · Break-even in 1.5 days
Start Free Trial — Save CHF 1’824/mo

14 days free · No credit card required

Pricing

Simple, transparent pricing

Start free, scale as you grow. All prices in CHF.

Free

Try EON with no credit card

Free
  • 50 AI Memories / month
  • 1 Project
  • Validator decisions not persisted
  • Community support
Start Free

Starter

For individuals and small projects

CHF 29/mo
  • 1,000 AI Memories
  • 10,000 API Calls/day
  • 500 MB Storage
  • 5 Projects
  • Email Support
Start Free Trial
Most Popular

Pro

For growing teams

CHF 99/mo
  • 10,000 AI Memories
  • 100,000 API Calls/day
  • 5 GB Storage
  • 50 Projects
  • Priority Support
  • Validator-decision export (JSON/CSV)
  • Browser Automation
Start Free Trial

Enterprise

For organisations with compliance needs

Contact us
  • Unlimited AI Memories
  • Dedicated tenant & SLA
  • DPA + Privacy Policy on request
  • AI-Act Article 12 audit export
  • Dedicated support
  • Custom Integrations
Talk to sales
FAQ

Common questions

Direct answers — no marketing fluff. If something is missing, write to hello@ai-developer.ch.

  • Built-in AI memories store everything raw. AxiomSeal validates each memory against 15 automated tests before storage — hallucinations, dogmatism, and unnatural phrasing get caught before they ever land. Your AI cites verified knowledge, not noise.
  • Yes. AxiomSeal exposes 26 tools through the Model Context Protocol (MCP) standard, including multimodal search (CLIP for images, Whisper for audio), bi-temporal queries, and cryptographic integrity checks. Any MCP-compatible IDE — Claude Code, Cursor, Windsurf, and others — connects in under 2 minutes via your bearer token.
  • Your data is stored in St. Gallen, Switzerland, on dedicated infrastructure. Each tenant has an isolated database — no shared tables. We are GDPR (EU) and DSG (CH) compliant. Export and full deletion are one-click operations.
  • Yes. The Starter plan (CHF 29/month) includes a 14-day free trial — no credit card required to start. You get full access to validation, semantic search, and MCP integration during the trial.
  • AxiomSeal's bi-temporal layer detects transition signals automatically. The old fact gets a valid_to timestamp, the new one becomes the current truth. You can still query "what was true on date X" — perfect audit trails without losing history.
  • Always. Full data export to JSONL or CSV is available on every tier. No vendor lock-in — your knowledge stays portable.
  • X-Ethics is a logical-consistency framework that scores each memory against truth-preserving axioms before storage. It catches contradictions, dogmatic absolutes, and hallucinated facts at write-time, not at read-time. Available on all tiers.
  • Both. The MCP server is the native path for IDEs. A REST API is available on Business and Enterprise tiers for custom integrations.
Still unsure? See how AxiomSeal compares to Mem, Letta, and others.

Start storing knowledge you can trust

Stop re-explaining your codebase every session. Give your AI a memory it can trust.

Start Free TrialRead the docs

No credit card required