Support
A contact page handles demo bookings and direct enquiries. The how it works, methodology and security pages document the runtime, and a showcase presents live examples.
Learn moreMnemom
Signed proof that your AI agents stayed inside their boundaries
Advanced
What is Mnemom?
Mnemom is an analytics tool. Mnemom governs AI agents before they act: Alignment Cards declare signed boundaries, AIP verifies reasoning in flight, AEGIS screens model calls at the gateway, and each run produces an Ed25519-signed record plus a public Trust Rating for auditors.
Mnemom is an integrity and governance layer for AI agents, positioned by its makers as the trust plane for the agentic internet. Rather than logging what an agent did after the fact, it sets the operating boundaries first and checks every decision against them while the agent runs. The core primitive is an Alignment Card, a versioned, signed document in plain terms that states what a given agent may and may not do. Each decision is then verified in flight against that card, so drift and policy breaks surface immediately instead of during a post-mortem review. The runtime is organised into five layers. AAP handles intent declaration through the Alignment Card. AIP verifies reasoning while it executes. CLPI governs and anchors permissions. Mnemom AEGIS screens every model call that passes through the gateway. A final layer computes Trust Ratings and coherence checks. The output of the whole chain is a tamper-evident record signed with Ed25519, plus a portable Trust Rating that can be described as a bond rating for agents rather than a star rating. Compliance bundles can be exported and handed to boards, auditors or regulators. The tool is aimed at teams deploying autonomous agents in regulated or high-stakes settings: security and compliance functions, AI platform owners, and engineering groups that need to demonstrate bounded behaviour to outside reviewers. Installation is quoted at under five minutes and the code is released under Apache 2.0, which suggests a self-hosted, developer-first deployment rather than a no-code console. In practice a user defines the card, routes agent traffic through the gateway, and receives signed evidence for each run. Among alternatives, Mnemom sits closer to policy enforcement and audit infrastructure than to observability dashboards or prompt monitoring tools. Those products typically report on outcomes; Mnemom intercepts before the action executes and produces cryptographic receipts. Buyers who only want traces and cost analytics will find it heavier than needed, while those who must prove containment to a third party are the intended audience.
How do you use Mnemom?
1Get started with an account
Create an account from the signup page to access the runtime. For a guided walkthrough before committing, the contact page books a demo with the team.2Read the five layer model
The how it works page explains the declare, verify and sign sequence, showing where each layer sits in a single agent run before you configure anything.3Draft your first Alignment Card
Write down what the agent may and may not do in plain, versioned terms and sign it. This card becomes the reference every runtime decision is checked against.4Route model calls through AEGIS
Point agent traffic at the Mnemom gateway so every model call is screened and verified in flight against the card you signed.5Review signed evidence and ratings
Inspect the Ed25519 signed receipts and the Trust Rating produced for each run, then export a compliance bundle when an auditor or board needs the record.6See a working deployment
The showcase page presents live examples of the stack in operation, useful for judging how the evidence output looks in a real review.
Pros and cons
Pros
- Checks policy before an action executes rather than reconstructing events afterwardsAI
- Ed25519 signed evidence and exportable compliance bundles suit audit and regulatory reviewAI
- Alignment Cards express boundaries in plain, versioned terms, which non-engineers can readAI
- Apache 2.0 licence and a quoted install time under five minutes lower adoption frictionAI
- Public, portable Trust Rating gives outside parties a verifiable signal about an agentAI
Cons
- Purpose-built for agent governance, so teams wanting simple tracing or cost dashboards will find it overbuiltAI
- Five interacting layers (AAP, AIP, CLPI, AEGIS, ratings) demand real setup effort before value appearsAI
- Trust Rating only matters if counterparties recognise the methodology, which is not yet an industry standardAI
- No pricing is published on the homepage, so budgeting requires a sales conversationAI
- Homepage gives no detail on which model providers or gateways AEGIS currently supportsAI
How much does Mnemom cost?
Features
Alignment Cards declare signed operating boundaries; AIP verifies reasoning in flight; CLPI governs and anchors permissions; Mnemom AEGIS screens every model call through the gateway; Trust Ratings and coherence scoring give agents a public, portable reputation. Every step produces an Ed25519 signed, tamper-evident record, and compliance bundles can be exported for boards, auditors and regulators.
Learn moreFrequently asked questions about Mnemom
How do you use Mnemom?
The walkthrough on this page covers 6 steps: 1. Get started with an account 2. Read the five layer model 3. Draft your first Alignment Card 4. Route model calls through AEGIS 5. Review signed evidence and ratings 6. See a working deployment.What platforms does Mnemom support?
Mnemom is available on Web App.What are the limitations of Mnemom?
Purpose-built for agent governance, so teams wanting simple tracing or cost dashboards will find it overbuilt. Five interacting layers (AAP, AIP, CLPI, AEGIS, ratings) demand real setup effort before value appears. Trust Rating only matters if counterparties recognise the methodology, which is not yet an industry standard.
Status
StatusActive
Views0
Outbound clicks0
Added10/21/2026
Platforms
Web App
Pricing
Paid
