Platform

The assurance engine.

Zansn Labs is an Australian deep-tech company building independent agent assurance. The assurance engine instruments any agent run and turns it into portable, verifiable evidence that an autonomous AI system behaved within bounds, evidence built to be checked independently, not taken on trust from the vendor whose agents are acting.

Scope
Reliability, observability, privacy.
Mechanism
Repeated runs, traces, evidence.
Outcome
Reviewable deployment decisions.
System view
01Research question
02Technical uncertainty
03Experiment protocol
04Repeated runs
05Artifacts and traces
06Metrics and failures
07Evidence graph
08Human review
09Decision record
10Archive evidence
Verifiable Evidence

Every evidence pack can be signed (Ed25519, zansn.signed-evidence-pack.v1) and checked offline by an auditor or insurer with the open-source zansn-verify CLI, no Zansn account or network call required.

Evaluation Harness

Runs repeated, seeded benchmark loops over domain tasks and records reliability, failure severity, and recovery quality.

Agent Trace Layer

Captures tool calls, policy checks, retries, state mutations, and failure attribution without depending on generic logs alone.

Agent Security Harness

Tests prompt injection, tool misuse, excessive agency, permission boundary failures, unsafe retries, and egress attempts before agents are connected to real systems.

Cost & Latency Profiler

Measures token cost, runtime cost, route-level latency, and the reliability frontier across workflow variants.

Privacy Boundary Tester

Checks where sensitive data can travel and flags unauthorized egress across prompts, traces, caches, tools, and APIs.

Evidence Graph

Connects hypotheses, experiments, runs, artifacts, metrics, failures, limitations, and decisions into a lineage record.

Deployment Readiness

Turns measured signals into an internal review aid for human decision-making, not a certification claim.

Local and cloud split
Local

Private inference, repeated runs, data-sensitive tests, and controlled baselines.

Cloud

Managed model comparison, scale tests, shared artifacts, and prototype APIs.

What is implemented
  • Four published, open-source framework adapters (Vercel AI SDK, LangChain, OpenAI Agents, MCP) that turn any agent run into a portable, signable evidence pack.
  • Public research library with note pages, references, diagrams, code blocks, and artifact links.
  • Contact flow for technical enquiries and project scoping.
  • Evidence graph visuals and readiness-oriented research tracks.
  • Local-first site architecture with static routes and searchable notes.
Developer Starter Kit

Start measuring agent behaviour locally

zansn is a local-first npm package that turns any agent run into a portable evidence pack. It runs without a Zansn account and transmits nothing, the open-source entry point to independent agent assurance.

What each evidence pack captures
Task successAgent-aware tracesTool-call evidencePolicy & privacy findingsCost & latencyDeployment-readiness signals
View on npmFree, local, vendor-neutral, and independently verifiable today: every pack can be signed and checked offline with no Zansn account. The hosted evidence cloud is in progress; independent attestation comes later.
evaluate.tstypescript
1import { ZansnClient } from "zansn";
2 
3const zansn = new ZansnClient({
4  defaultPolicy: {
5    allowedTools: ["lookupCustomer"],
6    blockedTools: ["deleteCustomer"],
7    maxEstimatedCostUsd: 0.05,
8    maxLatencyMs: 5_000,
9  },
10});
11 
12// Run a task through the assurance harness and get back
13// a portable, reviewable evidence pack.
14const evidence = await zansn.runEvaluation({
15  projectId: "ZAN-RD-001",
16  tasks: [{ id: "support-001", expected: "active" }],
17  execute: myAgent,
18});
19 
20// readiness → "ready" | "review" | "blocked"
21console.log(evidence.summary.readiness);
Works with your agent stack

Assurance for agents you already build

Already building with a popular agent framework? Add one line and the run you already have becomes a portable evidence pack, with the same policy, privacy, cost, and readiness checks. One adapter per framework, all open source.

@zansn/ai-sdk
with-assurance.tstypescript
1import { generateText } from "ai";
2import { createZansnCollector } from "@zansn/ai-sdk";
3 
4const zansn = createZansnCollector({ projectId: "support-agent" });
5 
6const result = await generateText({
7  model,
8  prompt,
9  onStepFinish: zansn.onStepFinish, // capture the run
10});
11 
12const pack = await zansn.finalize({ output: result.text });
13// pack.summary.readiness → "ready" | "review" | "blocked"