Audience: Technical leads, compliance stakeholders, and executive sponsors participating in an SWT3 pilot evaluation. Print this page, check the boxes, and hand it to your implementation team.
1. Purpose
This checklist covers every step from signed pilot agreement to your first verified SWT3 Witness Anchor. Print it, check the boxes, hand it to your technical lead. Total time: under 48 hours. For the full narrative explanation of the pilot program, evaluation criteria, and expected outcomes, see the Pilot Program Guide.
2. Phase 1: Legal and Administrative (Before Day 1)
Complete these items before any technical work begins. Contact engineering@tenovaai.com with any questions.
- Pilot Evaluation Agreement signed by both parties
- Mutual NDA in effect (included in the Pilot Agreement)
- Invoice issued ($10,000 USD, net-30)
- Kickoff session scheduled (60 minutes, remote)
- Technical lead and compliance stakeholder identified
- AI system for evaluation identified (one system per pilot)
3. Phase 2: Technical Prerequisites (Day 1)
Confirm these items with your technical lead before installing the SDK.
- Python 3.9+ or Node.js 18+ available in target environment
- Network access to
sovereign.tenova.ioon port 443 confirmed - AI provider API key available (OpenAI, Anthropic, Bedrock, vLLM, Ollama, etc.)
- Clearing level selected (see table below)
Clearing Level Reference
| Level | Name | What is recorded | Recommendation |
|---|---|---|---|
| 0 | Analytics | Hashes, factors, model name, provider | Internal analytics, no regulatory constraint |
| 1 | Standard | Hashes, factors, model name | RECOMMENDED for most pilots |
| 2 | Sensitive | Hashes and factors only | Healthcare, legal, PII-adjacent workloads |
| 3 | Classified | Numeric factors only, model name hashed | Defense, classified environments |
Network Verification
Run this command to confirm connectivity:
curl -s https://sovereign.tenova.io/api/v1/health | python3 -m json.tool
Expected output includes "status": "healthy" and a recent lastScan timestamp. If this command fails, check firewall rules for outbound HTTPS (port 443) to sovereign.tenova.io.
4. Phase 3: Account and SDK Setup (Day 1)
- Welcome email received with tenant ID
- Logged into sovereign.tenova.io and confirmed tenant ID in dashboard header
- API key copied and stored securely (shown once at signup, not recoverable)
- SDK installed (choose one):
Python
pip install swt3-ai
Verify installation:
python -c "import swt3_ai; print(swt3_ai.__version__)"
TypeScript / Node.js
npm install @tenova/swt3-ai
Verify installation:
node -e "const s = require('@tenova/swt3-ai'); console.log('swt3-ai installed')"
- SDK version verified (should be 0.6.5 or later)
5. Phase 4: First Witness (Day 1-2)
- AI client wrapped with Witness (3 lines of code -- see example below)
- First inference executed through the wrapped client
- Anchor appeared on AI Witness dashboard page (within 30 seconds)
- Anchor format verified: starts with
SWT3-followed by tier/provider/procedure/verdict/epoch/hash
Python Example (OpenAI)
from openai import OpenAI
from swt3_ai import Witness
witness = Witness(
endpoint="https://sovereign.tenova.io",
api_key="YOUR_KEY",
tenant_id="YOUR_TENANT",
clearing_level=1
)
client = witness.wrap(OpenAI())
# Every inference through this client now produces a witness anchor
response = client.chat.completions.create(
model="gpt-4o",
messages=[{"role": "user", "content": "Summarize our Q3 risk report."}]
)
TypeScript Example (OpenAI)
import OpenAI from "openai";
import { Witness } from "@tenova/swt3-ai";
const witness = new Witness({
endpoint: "https://sovereign.tenova.io",
apiKey: "YOUR_KEY",
tenantId: "YOUR_TENANT",
clearingLevel: 1,
});
const client = witness.wrap(new OpenAI());
const response = await client.chat.completions.create({
model: "gpt-4o",
messages: [{ role: "user", content: "Summarize our Q3 risk report." }],
});
wrap() call returns a proxy that intercepts every inference. It computes a SHA-256 fingerprint, assigns the appropriate UCT procedure (AI-INF.1), records the clearing-level-appropriate evidence factors, and transmits the SWT3 Witness Anchor to the ledger. Your AI provider call is unmodified -- same latency, same response, same error handling.
6. Phase 5: Verification (Day 2)
- Navigate to sovereign.tenova.io/verify
- Paste the SWT3 anchor string from the dashboard
- Verified status confirmed (green checkmark, fingerprint decomposition shown)
- Share the verification URL with your compliance stakeholder to confirm independent access
The verification page runs entirely in your browser using client-side SHA-256 computation. No inference content is transmitted during verification. Your compliance stakeholder can verify any anchor independently without a login.
7. You Are Ready When
- At least one anchor is verified on the public verification page
- Your compliance stakeholder can access the dashboard and see live evidence
- You have selected a clearing level appropriate for your data sensitivity
- Kickoff session is on the calendar
Once these four criteria are met, you are ready for the pilot evaluation sessions. The kickoff covers dashboard orientation, clearing level confirmation, and evidence mapping to your target framework. See the Pilot Program Guide for the full session schedule.
8. Stakeholder Handoff Matrix
Use this table to assign responsibilities across your team. Each role has distinct tasks -- overlap is intentional for the compliance stakeholder.
| Role | Responsibilities | When |
|---|---|---|
| Legal | Signs Pilot Agreement, reviews DPA and clearing addendum | Before Day 1 |
| Technical Lead | Installs SDK, wraps client, verifies first anchor | Day 1-2 |
| Compliance / Assessment | Reviews dashboard, selects clearing level, attends all 3 sessions | Day 1 through Day 10 |
| Executive Sponsor | Approves budget, attends findings session | Day 8-10 |
9. Related Documents
- Pilot Program Guide -- full narrative overview, session schedule, evaluation criteria
- Notified Body Pilot Package -- scope, deliverables, and pricing for NB evaluations
- NB Pilot Scope -- article-by-article coverage matrix and evidence boundaries
- Pilot FAQ -- common questions about timelines, pricing, data handling, and outcomes
- Pilot Evaluation Agreement -- binding terms and conditions
- AI Witness Quickstart -- SDK installation and first inference walkthrough
- Clearing Addendum -- data processing terms by clearing level
- SDK Documentation -- full API reference for Python and TypeScript