What each panel actually shows you, and how much of it is retrieved fact versus the AI's own read — worth knowing before you start combining them.
Live Price Ticker & Heatmap
Where prices are right now and what just moved, at a glance.
Real, retrieved data.
Price Chart
The actual historical path of one asset over 7D/30D/90D/1Y.
Real, retrieved data — the one place you are looking at a genuine chart, not a narrated one.
Prediction Window
A bull / base / bear scenario for an asset over a chosen horizon, with probabilities and reasoning.
Real price as the anchor; the scenario itself is the AI's judgment call, not a computed forecast.
Correlation Matrix
How strongly pairs of assets have actually moved together over your chosen lookback window.
Genuinely computed coefficients (real Pearson correlation from stored history); the written read of what it means is the AI's. Correlation is never treated as causation here — the AI is instructed not to claim it.
Futures Curve
The term structure for an asset — whether the market is pricing contango or backwardation.
Curve shape is real; any narrative about why it looks that way is the AI's interpretation.
News Digest
"Energy news you might have missed," filtered to your topics and tagged bullish / bearish / neutral.
Headlines are real (pulled from verified RSS sources); the sentiment tag on each is the AI's read.
Geopolitics Layer
Structured conflict, sanctions, and "reduced relations" events, scoped to energy-strategic countries.
Sourced from the GDELT Project's public event database — real recorded events, not AI-invented ones; the impact/hedge commentary on top is the AI's.
Compute & Tech Layer
AI training-compute growth, hyperscaler capex, data-center power draw, and chip export-control changes — the infrastructure behind AI's own energy appetite.
Figures come from Epoch AI, SEC filings, and trade.gov; the "so what does this mean for gas/power demand" narrative is the AI's.
Live Map
Physical energy infrastructure — power plants, pipelines, LNG terminals, refineries — plotted globally; click a node for an AI read on it.
Node locations are real (EIA/Global Energy Monitor data); the node-level commentary you get on click is AI-generated.
Africa Dashboard
Nigeria Bonny Light plus wider pan-African energy context in one panel.
Worth knowing: Bonny Light is tracked as a Brent proxy, not its own direct feed — so a "Bonny move" is really a Brent move with regional context layered on.
Chat Terminal
Open-ended energy Q&A — ask it anything, including things no single widget covers (e.g. "how do export controls affect gas demand").
Every answer requires a live tool call first — the AI is not allowed to answer from memory alone.
Analysis Hub
Takes two or more analyses you've already had (from any of the above) and overlays them into one consolidated read.
Your own saved conversations are real; the synthesis across them is the AI reconciling — or flagging a conflict between — what you fed it.
The real value comes from overlaying tools, not reading one in isolation. Here is what to combine for a given goal, in what order, and what the combination actually tells you that any single panel wouldn't.
Spot a hotspot before the headlines catch up
- Scan the Live Map for infrastructure clustered near a region — a strait, a pipeline corridor, a refining hub.
- Overlay it with the Geopolitics Layer, filtered to that country: is verbal or material conflict activity actually rising there in the last few days, or is the map just quiet?
- Cross-check with the News Digest, scoped to the same region/topic: has mainstream coverage caught up yet, or are you looking at this early?
What you get: A map alone shows you where things are, not whether anything is happening there. Layering it with real conflict events turns "infrastructure exists here" into "infrastructure that matters is currently under strain here" — and the news check tells you whether that's already common knowledge or still ahead of the story.
Reality check: The map and the events feed are both real, retrieved records. Whether a given event is actually "energy-relevant enough to matter" is still your judgment call — treat this as a shortlist to investigate, not a finished conclusion.
Turn a hotspot into a market read
- Once you have a hotspot (see above), identify which asset it's strategically tied to — a Gulf chokepoint points to crude, a European pipeline node points to TTF gas.
- Open that asset's Price Chart: has price action already moved, or is it flat — meaning the market hasn't reacted yet?
- Check the Prediction Window for the AI's directional read given the same event.
- Run the Correlation Matrix across that asset and its close relatives (e.g. Brent, TTF Gas, EU ETS together). A hotspot that's genuinely moving markets tends to show up as several logically-linked assets moving together, not noise in just one.
What you get: This is the difference between "something is happening" and "something is happening that the market has or hasn't priced in yet." A flat chart plus a rising-tension event feed is a very different situation from a chart that has already moved.
Reality check: The price history and correlation coefficients are computed, real numbers. The prediction and the "is this priced in" read are the AI's interpretation — useful as a hypothesis to test against the next few days of real data, not a guarantee.
Sanity-check a sudden price move
- Start at the Heatmap: what actually moved, and by how much, across the board?
- Zoom into the Price Chart for the asset in question — is this a sharp spike or a trend that's been building for days?
- Run the Correlation Matrix: is the move isolated to this one asset, or systemic across the whole energy complex?
- Check the News Digest for a specific headline. A quiet news feed alongside an isolated move often points to positioning or technical trading rather than a fundamental cause.
What you get: Isolated-plus-news-driven, systemic-plus-quiet, isolated-plus-quiet — each combination points you toward a different explanation, and toward a different question worth asking next.
Reality check: All four data points here are real and retrieved. This playbook is deliberately data-only — it is the one worth trusting most at face value.
Build a directional view you can defend
- Start with the Prediction Window's bull/base/bear scenarios as your working hypothesis, not your conclusion.
- Check the Futures Curve: does the term structure (contango vs backwardation) agree with that direction, or point the other way?
- Run the Correlation Matrix for confirmation — are related assets moving in a way that's consistent with the thesis?
- Look for a fundamental catalyst in the Geopolitics Layer rather than relying on chart pattern alone.
What you get: A thesis that survives three independent checks — curve shape, cross-asset correlation, and a real catalyst — is a much stronger starting point than a single prediction taken at face value.
Reality check: This is the playbook to hand to the Analysis Hub (see below) — export the Prediction Window chat and the Correlation Matrix chat as two threads and ask the Hub explicitly whether they agree.
Track the AI-compute-demand story as an energy driver
- Ask the Chat Terminal directly — "how are hyperscaler capex and chip export controls affecting AI compute capacity?" — to pull the current training-compute, capex, and power-draw trend.
- Run the Correlation Matrix between Natural Gas or Uranium and the broader energy complex to see whether power-hungry AI buildout is already visible in that pricing, or still a slow-burn story.
- If data-center coordinates have been imported, check the Live Map for physical clustering near existing gas or grid infrastructure.
What you get: This is a genuinely new-ish demand driver for the energy market, and it moves on a different clock than geopolitics — capex and training-compute trends shift quarterly, not daily. Checking it occasionally rather than constantly is usually the right cadence.
Reality check: The underlying compute/capex/export-control figures are real, sourced numbers. Whether they are "already priced into gas" is an open, AI-interpreted question — this is one of the more speculative playbooks here.
Nigeria / Bonny Light focus
- Start with the Africa Dashboard for the on-the-ground snapshot and pan-African context.
- Filter the Geopolitics Layer to Nigeria and neighbouring energy-strategic states for anything moving beneath the surface.
- Cross-reference against Brent in the Correlation Matrix — since Bonny Light is tracked as a Brent proxy, a Bonny "move" is best read as a Brent move with a regional story layered on top, and Brent's own chart is the more reliable anchor.
What you get: This keeps a regional focus from turning into a false sense of a separate, independently-moving market — Bonny Light's price behaviour is, structurally, Brent's.
Reality check: Worth restating: there is currently no direct licensed Bonny Light feed in this plugin. Treat the Africa Dashboard's price reads as "Brent, with Nigeria context," not as an independently verified regional price.
Analysis Hub overlays two or more of your saved widget conversations into one AI-synthesised read — shared themes, contradictions, and a consolidated view. Here is how to actually use it as a synthesis layer, not just a chat archive.
Have a real conversation first
Open any widget in fullscreen and use its Analyze chat — this is a scoped conversation with real back-and-forth memory about that widget's live data, not a one-shot summary. Push it until you have an actual conclusion, not just the widget's default output.
Export it with a name that means something later
"WTI bull case — Hormuz tension" is far more useful three days from now than "chat 1." You'll be picking these out of a list.
Do it again from a different angle
The Hub needs two or more threads, and it works best when they genuinely could agree or disagree — a price/technical read alongside a geopolitical read, for instance, rather than two similar takes on the same widget.
Select them in the Hub and ask a specific question
"Do these support a bullish or bearish 30-day view, and where do they actually disagree?" gets a much sharper synthesis than leaving it open-ended.
Treat the synthesis as your next question, not your final answer
It has its own follow-up chat on top of the combined read — push back on it, ask what would change its mind, and keep going.
Combinations worth trying
Questions about interpreting what you're looking at, grouped by topic.
Reading the signals
Neither on its own — that disagreement is exactly the signal to dig further, not to average them out. Pull both into the Analysis Hub and ask it directly where they actually diverge; often one is reacting to a headline the other hasn't weighted yet.
No, and the AI is specifically instructed not to claim that. Two assets can move together because of a shared third cause (a macro shock, a regional event) without either driving the other. Use it to confirm a move is systemic, not to assign blame.
Check whether the asset it's tied to has actually moved (Price Chart) and whether related assets are moving with it (Correlation Matrix). Infrastructure existing somewhere tense is not the same as that tension being priced in yet.
Treat it as a sharper starting hypothesis, not a verdict — it is still the AI's reconciliation of conversations you fed it, not an independent check. Use its follow-up chat to stress-test it before treating it as settled.
Specific tools
It's tracked as a Brent proxy rather than a direct licensed feed. Read it as "Brent, with Nigeria-specific context added" rather than an independently-verified regional price.
RSI-14 needs at least 15 stored data points; correlation needs at least 5 overlapping points per asset pair. Below that, the figure comes back as "still accumulating" rather than a guess — if you see that message, the honest answer is to wait rather than read into an early number.
The capex, training-compute, and power-draw figures are sourced and real. Whether that translates into gas or power prices moving is the AI's read of a genuinely open, still-developing question — worth treating as informed speculation, not settled fact.