Labs

Ideas you can open and use

A lab is a working piece of an idea. Each one says what data it runs on, whether you can open it today, and which idea it belongs to.

• Sign-in

  Market Shards

  Every S&P 500 company drawn in shards instead of candles, its tree beside its chart. Click a shard for the nearest shapes in other companies.

  S&P 500 daily prices

  Idea: Irregular series as trees of shards

• Public

  Earnings Lab

  S&P 500 earnings in an ontology-typed knowledge graph, with the passage behind every fact and an agent that cites it.

  Live data

  Idea: Knowledge an agent can cite

• Public

  Structure in a series

  A series with days missing, cut at its turning points and grouped into a tree of shapes.

  Synthetic

  Idea: Irregular series as trees of shards

• Pre-launch

  GIFS Segmentation API

  Send a series, get its tree of shards back. Accounts, tiers and credits are built.

  Your data

  Idea: Irregular series as trees of shards

• Not yet hosted

  2D layer viewer

  The tree of shapes over a real street photograph, layer by layer.

  Public photograph

  Idea: From 1D series to 2D images

• Not yet hosted

  Shape widgets

  Histories drawn from small GIFS files at the detail a widget has room for.

  Daily prices

  Idea: Shape as a format

• Public

  AI Impact Assessment

  An interactive treemap of a capability model, each capability scored for AI impact by an LLM.

  LLM-scored capability model

  The method behind the work: open source

• Public

  Foresight, Sentinel and Essence

  Three concept figures: forecasting, anomalies and patterns by matching shapes against the past.

  Synthetic

  Idea: Reading new data by precedent

Interactive figures

Each figure is a working dashboard that also runs on its own, and says what data it shows.

• Figure

  Essence: patterns by precedent

  Show me every time any asset looked like this, and what happened next. Pick a shape on any signal and see its nearest precedents across a fleet, at other time scales too, each with the work order or outcome that followed.

  synthetic Six fictional assets with recurring shapes and made-up maintenance records. An illustrative concept, not a measured result.

• Figure

  Foresight: forecasting by analogy

  Find the past windows whose shape is nearest the latest window and lay what followed them over the present, as a fan with a median and a band. Change the window and the number of analogues, or drag now back in time to see what actually happened.

  synthetic One seeded series built from recurring shapes. An illustrative concept, not a forecast or a measured result.

• Figure

  Sentinel: anomalies by shape

  A transformer, a pump, a flow meter and a wind turbine, each scored by its distance from its nearest normal shape, on one axis with one threshold. Replay three weeks to see one threshold catch slow faults the per-asset limits miss, without the false alarm a limit raises, and open any alert beside the normal behaviour it was compared with.

  synthetic Four fictional assets with faults injected at known times. An illustrative concept, not utility data or a measured result.

• Figure

  A tree of shapes

  How a historian archive becomes a tree of shapes: the segments the historian kept are the leaves, channel breaks group them into larger shapes, and each new segment is compared with trees from normal operation, so a shape unlike normal stands out while values are in range.

  synthetic An illustration of the method on made-up data. It makes no measured claim.

• Figure

  Structure in an irregular series

  An irregular daily series stored as a hierarchy of shapes, coarse first, and drawn with the earnings lab's price chart. Drag the tolerance to see how much structure each level keeps.

  synthetic A seeded random walk on trading days, with days missing. Illustrative, not market data.