Learn

Learn how KrisVen makes AI decisions provable.

Short videos, learning paths and feature guides, including the four KrisVen tools coming to AWS Marketplace. Each feature is labelled with its current status, so you know what you can try today.

Start here

Two short videos

Watch one of these first. Both use synthetic data.

What KrisVen does

A tour of governed memory, expert reasoning, human review and the Decision Ledger.

KrisVen on AWS Marketplace · under a minute

The four tools, how onboarding works and answers to common concerns.

Learning paths

Pick the path for your role

Each path takes about 20 minutes and uses only the pages on this site.

Business leaders

Why decisions need evidence

Where AI decisions create cost and risk, and what changes once each one can be proven.

  1. No Regulator Is Asking. Your Customers Are.
  2. Where Fintech Value Creation Actually Happens
  3. Check your AI Trust Score
Risk & compliance

What an examiner will ask

How approved knowledge, human review and sealed records answer the question months later.

  1. Six Months Later: The Question Sam Couldn’t Answer
  2. Why Commercial-Lending AI Needs an Evidence Layer
  3. Security
Platform engineers

How to connect your agents

The architecture, the MCP tools and how an agent reaches them through your gateway.

  1. Architecture
  2. Getting started on AWS Marketplace
  3. Evidence without moving your data
Feature guides

What KrisVen does, feature by feature

Status labels: Sandbox you can try it in a KrisBox sandbox today. Coming soon built and tested, waiting on an external listing. Planned designed, not built yet.

Governed memory

Sandbox

Agents read only knowledge that a person has approved, and every answer cites its source.

  • Policies, thresholds and precedents are versioned
  • Each organization’s memory is isolated from the others
  • Agent writes stay drafts until someone approves them
Try it in KrisBox →

Expert reasoning and human review

Sandbox

Domain checks run before a decision goes out, and anything outside policy goes to a person.

  • Reviewers see the evidence the agent used
  • Their approval or override is recorded with the decision
See the domains →

Decision Ledger

Sandbox

Each decision is sealed with an Ed25519 signature over what it relied on, and anyone with the public key can verify it later.

  • Replay shows what the agent saw at the time
  • Offline verification needs no KrisVen account
  • If something changed, verification names the field that changed
How sealing works →

Controlled actions

Sandbox

Agents act only within limits you set: which tools they may use and when a person must approve.

  • Every action carries the decision that authorized it
  • Out-of-policy requests are refused with a reason
See the architecture →

Agent commerce

Sandbox · test payments

Agents from other companies can discover KrisVen and transact with it over A2A, MCP or REST, using test payments only.

  • Partner agents get scoped credentials
  • Payment mandates follow the AP2 protocol
  • Each purchase is sealed like any other decision

KrisVen MCP tools on AWS Marketplace

Coming soon

Four tools in one remote MCP server, billed per operation through your AWS account.

  • krisven_read: approved context, with sources
  • krisven_write: drafts that wait for a person to approve
  • krisven_seal: a signed record of a decision and its evidence
  • krisven_verify: checks a record and names any change
Getting started →

Fingerprint-only evidence

New

Seal can work from SHA-256 fingerprints of your documents instead of the documents themselves.

  • Your files stay in your own storage
  • Verification compares fingerprints, so any edit shows up
How it works →

Amazon Bedrock AgentCore Gateway

Coming soon

Add KrisVen to your gateway as an MCP target, and every agent behind it sees the four tools.

  • Machine-to-machine OAuth, with the client secret kept in your Secrets Manager
  • Works with other MCP clients too

Deployment in your AWS account

Planned

A container version of KrisVen that runs inside your own AWS account. It is designed but not built.

AI Trust Score

Available now

A short self-assessment showing where your AI decisions are hard to prove today.

Get your score →
AWS Marketplace · coming soon

Getting started on AWS Marketplace

The listing is not live yet. This is how onboarding will work once it is published.

  1. Subscribe

    Find KrisVen in AWS Marketplace and subscribe. Charges per operation appear on your AWS bill.

    Who: your AWS buyer or admin
  2. Link your organization

    AWS sends you to KrisVen Studio, where you link the subscription to your organization and choose which tools to enable.

    Who: your KrisVen admin
  3. Receive credentials

    An OAuth client is delivered to Secrets Manager in your own AWS account. It is never emailed or shown on screen.

    Who: automatic
  4. Connect your gateway

    Add KrisVen as an MCP target in AgentCore Gateway, or in any MCP client, using those credentials.

    Who: your platform engineer
  5. Load and approve one workflow

    Load the policy, thresholds and precedents for one workflow, then approve them so agents can read them.

    Who: your domain owner
  6. Seal and verify your first decision

    Call krisven_seal on a decision, then run krisven_verify on it, the way an auditor would.

    Who: your team
Your data

Evidence without moving your data

KrisVen needs to prove what a decision relied on. It does not need a copy of your documents to do that.

YOUR AWS ACCOUNT Documents stay in your storage Your agent hashes files locally Gateway AgentCore or MCP Secrets Manager holds the OAuth client secret queries · fingerprints context · signed records KrisVen service approved knowledge Decision Ledger

The knowledge you choose to load into governed memory, such as approved policies and thresholds, is stored by KrisVen. Supporting documents for a decision can stay with you as fingerprints.

Stays with you

  • Source documents and customer records
  • The OAuth client secret
  • Access control and retention for your files

Sent to KrisVen

  • Knowledge you choose to load and approve
  • Tool calls from your agents
  • SHA-256 fingerprints of evidence
Glossary

Terms you will see

Governed memory
Knowledge an agent may use only after a person approves it, versioned and tied to its source.
Decision Ledger
The store of sealed decisions, each with what it relied on and who approved it.
Seal
A record of a decision signed with an Ed25519 key, so any later change can be detected.
Fingerprint
A SHA-256 hash of a document. It identifies the exact content but cannot be turned back into it. It is not a signature.
MCP
Model Context Protocol, an open standard that lets AI agents call tools on remote servers.
A2A
Agent-to-agent protocol, used when another company’s agent talks to KrisVen.
AgentCore Gateway
An Amazon Bedrock service that puts MCP tools behind one endpoint for your agents.
KrisBox
A KrisVen sandbox where you can try the features with synthetic data.
FAQ

Questions people ask

Is KrisVen on AWS Marketplace now?

Not yet. The listing is being prepared. This page will be updated when it is published.

How will billing work?

Pay per operation, with separate counts for Read, Write, Seal and Verify. Charges appear on your AWS bill. Prices will be shown on the listing.

Can an agent approve its own writes?

No. krisven_write creates a draft. Only a person in your organization can approve it into governed memory.

What happens if we cancel?

Your credentials stop working, and further calls are refused. You can still verify records you sealed earlier, offline, with the public key.

Do we need AgentCore Gateway?

No. Any MCP client that supports OAuth client credentials can call the tools. AgentCore Gateway is one convenient option on AWS.

Where can we try it today?

In a KrisBox sandbox with synthetic data. Ask for access.

More to read

Go deeper

Start with one workflow. Seal one decision. Verify it the way an auditor would.

Talk to us about a KrisBox sandbox, or get your AI Trust Score first.