AI-operated sweeps · human-owned judgement · every claim sourced · every miss recorded

Nownet · Grand Unified Thesis ·

State of the World — Signal Monitor

Twenty signals across five domains, graded against the tape rather than the narrative. One signal flashing is normal; three or four at once is the model starting to crack. A research monitor, not a trading system.

Cluster status

This is the reference instance — the monitor we run on our own signal set. A client's monitor looks like this with their signals, thresholds and tiers.
Encode your logic →

Countdown

Days from today to the dates on which the framework's assumptions are most likely to be tested.

Anchor values

Latest reading against the 1 June baseline (struck through). Data to 30 October 2026, the last trading day before this sweep.

The twenty signals

Confirmed, open, falsified

What the five months since June did to the thesis. Falsifications are recorded as falsifications, not reinterpreted as delay.

What to watch — next ninety days

DateEventWhat it tests

How this page is produced

This monitor is operated by an AI research pipeline under a fixed methodology, with a human owner setting the logic and signing off the judgement. The same platform can be geared to your organisation: what is watched, how it is graded, and how it is reported are all yours to define.

01 · Scope

A signal set with explicit thresholds, grouped into domains, encoded once as the organisation's logic.

02 · Research

Parallel strands sweep the period across every domain and leave dated, sourced notes with conflicts flagged.

03 · Verify

An independent pass checks the high-impact claims against primary sources before anything is graded.

04 · Grade

Each signal is scored against its threshold; the cluster is computed; the ledger is updated against the previous cycle's own expectations.

05 · Publish

A dated edition, this dashboard and a machine-readable data file — earlier cycles never rewritten.

Bespoke tailoring. The methodology is fixed; everything it is applied to is configurable — your own signals and thresholds, your structural tests and cluster tiers, your cadence, your formats, audiences and house style, your guardrails. A geopolitical-investment monitor, a supplier-risk monitor and a regulatory monitor are the same engine with different logic.
About the platform →

Method. Status changes on sustained moves, not single prints. Partial activations count at half weight toward the cluster score; early warnings count toward the next tier only if they convert. The framework identifies the conditions under which conventional analysis becomes unreliable; it does not forecast timing and it is not investment advice. Companion to State of the World (Part XIII, 1 November 2026) and Connecting the Dots (Nownet, May 2026). Privacy.