Styxis
Company
Knowledge

Preparing a 2027 AI-RAN trial is not reporting a trial result today

Approve the site, platform versions and traffic conditions before the trial handoff.

There is useful work before a result

A planned commercial-network trial creates work that can begin now, but its future result cannot be reported in advance. The immediate decision is whether the preparation package is ready. Truthound can check a proposed readiness table for missing prerequisites and allowed stage labels. It cannot validate radio performance or establish that a network trial has occurred.

The planned Japan trial

Nokia and SoftBank announced on October 20, 2026 an agreement for an AI-RAN field trial in Japan in 2027. The release describes planned platform components and commercial traffic conditions. It is a dated preparation commitment, separate from laboratory work or another platform announcement. The intended trial remains a future event within that release's scope.

Make the handoff reproducible

For a handoff, identify the site, platform components, software versions, approved traffic window, approval state and trial stage. Give each prerequisite an owner. A version field filled with latest is not a reproducible configuration, and an intended traffic window is not evidence that the traffic was observed. Keep plans and observations in separate fields.

Check the prerequisites

For the field-trial readiness package prepared as a Polars DataFrame, Truthound Core Python SDK 3.3.4's NotNullValidator can detect absent required fields in this prepared package. An allowed-value rule can keep preparation, approved trial and completed observation distinct. These are checks on supplied records. A passing package does not authorize network work; the operator's approval and engineering procedure determine what can proceed.

A readiness-contract alternative

Soda Core 4.30.0 provides YAML-contract missing checks for supported data sources. If a governed readiness register already has such a contract, keep that operational boundary; use a local Truthound check for a deliberately prepared extract when appropriate. Soda v4 packages and Elastic License 2.0 need consideration. Neither validation route establishes successful radio behavior.

Name today's stop condition

Define an immediate stop condition: no confirmed approval, unresolved component version or missing permitted traffic window means the handoff remains preparation. Use a small synthetic package to verify that a completed-looking label cannot bypass a missing prerequisite. Avoid changing an unknown to approved merely to satisfy a schema.

Create observations after the event

Once an authorized trial actually happens, create a separately identified observation record with its own conditions and evidence. Compare it with the preparation assumptions without rewriting the original plan. This preserves what the agreement made possible today and what a future experiment may teach, while keeping promised deployment and measured behavior apart.

Evidence

Event checked
Subject checked
Modified