Core concepts

Sources, articles, and jobs

A source is a public page discovered by search or submitted by URL. Once registered, it appears as an article. A job records the work required to retrieve and analyze that article, including its status, attempts, extraction mode, and any failure reason.

Workspaces and the canonical graph

Articles, claims, and entities are processed once in the canonical graph. A personal workspace selects articles from that graph, and its companies, insights, and weights are calculated only from claims reachable through those articles. Removing a workspace does not delete shared intelligence.

Discovery agents

Discovery agents are scheduled research configurations attached to a workspace. A bounded shared executor runs due agents, globally deduplicates URLs, and associates resulting articles with the requesting workspace. Availability depends on plan.

Claims and context

A claim is a discrete assertion extracted from a source. Claims are linked to relevant actors and contexts, such as an event, policy, market, product, or supply-chain relationship. These links provide the evidence used to derive company insights.

Companies

Companies are the main subjects of impact analysis. The Companies view can include a name, ticker, sector, industry, country, impact weight, and insight count. Select a row to see the company’s positive and negative insights together with their reasoning.

Insights

An insight is TraceGraphs’ interpretation of how connected evidence may affect a company. It combines a direction, score, confidence, context, exposure, derivation method, and reasoning note. An insight is a research lead, not a verified prediction or recommendation.

Entity weights

Entities are named actors found across claims, such as organizations, people, locations, products, and institutions. Net impact summarizes how source-backed claims affect each entity within the selected time view. Impact exposure shows the total strength of those effects before positive and negative claims cancel. Claim and edge counts indicate how much graph evidence contributes to the scores.

• Positive and negative describe the direction of modeled impact on the entity.
• Larger absolute net impacts indicate a stronger modeled effect on the entity.
• Impact exposure measures effect strength, not intrinsic or financial value.
• Weights can change when new sources arrive or when the time influence setting changes.

Time influence

• Decay gives newer evidence more influence; half-life controls how quickly older evidence loses weight.
• Date range uses only evidence inside the dates you select.
• All time considers every available claim without reducing influence based on age.

Your time influence selection is remembered in this browser and is shared by the Insights, Companies detail, and Entity Weights views.