Seedlabs

GazeSteer: Attention-Driven Steering for Progressive Analyst Workstations

A tool that tracks where an analyst looks to automatically focus computer power on that specific data.

A workstation add-on that tracks where an analyst is looking (or dwelling) in a partial result and continuously re-steers the running computation toward that region's data subspace — compute follows attention, with no explicit lasso required.

Computer ScienceData Visualization and Analytics
Cognitive-Adaptive Visual AnalyticsSecurity operations centers: analysts reviewing progressive scans of security telemetry get the events matching the threat they are currently examining computed first.

Concept

Steering-by-example still asks the analyst to stop and explicitly select items. In hours-long exploratory sessions, attention itself is a signal: attention-aware visualization research shows a system can track which parts of a display the user is examining over time and feed that information back into the visualization. This product fuses the two capabilities for high-value analyst workstations: an eye tracker (with hover/dwell interaction telemetry as a fallback) identifies the items under sustained attention in the partial view; those items become a rolling "example set"; and the steering engine regenerates its relaxed priority query every few seconds so the computation queue keeps filling in wherever the analyst's focus actually is. The analyst never issues a steering command — the data they are examining simply completes first. Position it for settings where a tracker is a trivial cost relative to the compute being steered: investigation desks, threat hunting, and large-cohort research review.

Why now

The new paper proves the steering half: relaxed queries generated from a set of selected items considerably improve Precision and Recall for prioritizing unprocessed data versus random uniform sampling, and the approach needs no index structures — making it cheap enough to re-run continuously from a noisy signal like gaze. The older attention-aware visualization work [1] proves the perception half: a system can track what the user is looking at over time and react appropriately. Neither alone delivers implicit steering; together they define a buildable capability with a clear prototype path (tracker + progressive engine + rolling example set).

Written by GLM · 8 Oct 2026

AI assessment

Backed by 2 papers78

A high-utility integration of gaze-tracking and progressive compute that removes manual friction from data exploration, though it is a direct evolution of the existing SteerQueue concept.

Evidence strength
4/5
The idea synthesizes two distinct, proven research components—relaxed query steering and attention-aware visualization—into a logical pipeline.
Market pull
4/5
SOC analysts in high-stakes environments have a strong incentive to reduce 'time-to-insight' and the budget to support specialized hardware like Tobii trackers.
Novelty & moat
2/5
The core mechanism is nearly identical to SteerQueue, simply replacing a manual 'lasso' with an automated 'gaze' trigger.
Feasibility
4/5
The components (eye-trackers and progressive engines) already exist; the primary challenge is the integration of the rolling example set logic.
Wedge clarity
5/5
The focus on security telemetry in SOCs provides a sharp, high-value entry point with a clear user persona and urgent need.
Simplicity / focus
5/5
The product is a single, focused add-on with one clear function: automating the steering of compute based on visual attention.

Scored by Gemma 4 · 8 Oct 2026. Rated against a fixed rubric (evidence, market, novelty, feasibility, wedge, simplicity), relative to the closest ideas already in the catalog. A prior estimate to compare ideas before real-world signal arrives.

Persona discussion

AI personas trained on real people's expertise debate this idea as it evolves.

View the discussion →

Act on this idea

Ideas only matter if someone runs with them. Your message goes straight to the founder's inbox — nothing is stored on our servers.

Who benefits

  • Tobiicompany

    Tobii sells eye-tracking hardware and SDKs and actively seeks professional verticals beyond research; this gives its trackers a concrete workstation use case tied directly to compute savings on large jobs.

  • Splunkcompany

    SOC analysts run long searches over security telemetry; attention-steered progressive search would prioritize the events around whatever the analyst is scrutinizing, cutting the wait for relevant results.

  • CrowdStrikecompany

    Threat hunters at CrowdStrike query massive endpoint telemetry archives; implicit steering would keep compute focused on the indicator cluster under active investigation without breaking their workflow.

  • MITREorganization

    MITRE researches and evaluates SOC tooling and analyst workflows; an attention-steered progressive analytics prototype is a natural applied-research pilot in its lab and evaluation environments.

Research it builds on

  1. Steering-by-example for Progressive Visual Analytics
    Marius Hogräfer, Marco Angelini, Giuseppe Santucci et al. · 2022 · 10 citations
    All ideas from this paper →
  2. Attention-Aware Visualization: Tracking and Responding to User Perception Over Time
    Arvind Srinivasan, Johannes Ellemose, Peter Butcher et al. · 2024 · 9 citations
    All ideas from this paper →
Leave feedback
feasibility