Most risk frameworks assume the world behaves normally. This one watches for when it stops.
Correlations that held for decades break in days; the models that priced them keep producing confident answers. The platform's job is not to forecast outcomes but to detect the conditions under which forecasts become unreliable — and to say so in writing, with a record of what it expected and what it got wrong.
Connect, don't itemise
Events are read as expressions of underlying structural themes, not as a feed of headlines. A tariff, a treaty and a bond auction can be one story.
Grade against the tape
Every signal carries an explicit activation threshold. Status changes on sustained moves, never on a single print or a compelling narrative.
Count the cluster
One signal flashing is ordinary. Several flashing together is the model cracking. The cluster — and its composition — is the output.
Keep the ledger honest
Each cycle records what was confirmed, what is open and what was falsified. Contradictions are recorded as contradictions, not reinterpreted as delay.
How a sweep runs
An AI research team operates the cycle end to end under a fixed methodology; a human owner sets the logic and signs off the judgement. The platform was itself built this way — the reference instance's editions, dashboard and data series are produced by the pipeline you would be buying.
Scope
Your signal set, domains, thresholds and reporting rules, encoded once as the organisation's logic.
Research
Parallel research strands sweep the period across every domain, producing dated, sourced notes with conflicts flagged.
Verify
An independent pass checks the high-impact claims — leadership changes, rate decisions, treaties, status-changing figures — against primary sources.
Grade
Each signal is scored against its threshold; the cluster is computed; the verdict ledger is updated against the previous cycle's own expectations.
Publish
A dated edition appended to your series, a dashboard, and a machine-readable data file — nothing from earlier cycles rewritten.
Tailored to your organisation's logic
The methodology is fixed. Everything it is applied to is yours to define. The same engine that runs a geopolitical-investment monitor can run a supply-chain, country-risk, sector or regulatory monitor — bespoke to what you need reported, how, and to whom.
Your signal set
Your own indicators and their activation thresholds, grouped into the domains that matter to you — currency, credit, commodities, suppliers, jurisdictions, competitors, regulators.
Your structural tests
Which events count as structural rather than noise, how partial activations score, and where your cluster tiers sit — calibrated to your risk appetite.
Your format and voice
Board paper, desk note, one-page RAG summary or full edition; house style and tone; British or American English; the depth each audience needs.
Your cadence
Monthly, weekly, or triggered — a scheduled catalyst, a threshold breach, a date on your own calendar.
Your distribution
A private dashboard, an internal site, a data feed into your own tools, or a PDF to the people who need it. Public or private, at your discretion.
Your guardrails
The bias-discipline controls, the research-not-advice framing, and any compliance language your organisation requires, built into every output.
What you receive each cycle
The edition
A dated analytical document appended to your series — status, theatres that moved, the graded signal table, the verdict ledger and the forward calendar.
The dashboard
A live page: cluster status, countdown to the dates that test your assumptions, anchor values against the previous baseline, every signal with its evidence.
The data series
A versioned, machine-readable file per cycle — signals, grades, verdicts, anchors — so history is queryable and your own systems can read it.
The research record
The sourced notes behind every grade, so any judgement can be traced to the evidence and the date it was gathered.
What it is not
Not a trading system. The platform produces understanding, not trade signals, and says so in every output. Applied to timing decisions it has, by its own author's account, underperformed a passive index.
Not a forecast. It identifies the conditions under which forecasts become unreliable. It does not predict when a structural shift produces visible consequences.
Not a news feed. Most events are noise. The platform's value is the structured weighting that decides which are not.
Not neutral. Every framework reflects its priors. Yours will reflect your organisation's — made explicit, so they can be tested.
Engagement
Three ways in. Pricing is on request and depends on the number of signals, domains and audiences; a pilot is scoped in a single working session.
Pilot
- Your signal set and thresholds encoded
- Three monthly sweeps with edition, dashboard and data file
- Private site; reference-instance design
- Review session after each cycle
Desk
- Everything in Pilot, on a standing cadence
- Event-driven re-sweeps on your triggers
- Your house style, formats and audiences
- Data feed into your own tools
- Quarterly review of the signal set itself
Enterprise
- Multiple monitors across functions or regions
- Cross-monitor cluster view
- Private deployment inside your environment
- Compliance language and governance controls
- Named analyst owner
To scope a pilot, send a line describing what your organisation needs watched and who needs to read it. A first conversation is enough to draft a signal set.
Email: phil@nownetsecurity.com
Built by Nownet. The platform grew out of a working research practice — the reference monitor and its white paper series have run since March 2026 — and the methodology is published in Connecting the Dots (May 2026). The operating pipeline, the signal library and the grading logic are proprietary.