SpyderBot Documentation
Agent Count
Agent Count determines the observation depth of a Prompt Explorer investigation.
Overview
Rather than relying on a single AI interaction, Prompt Explorer coordinates multiple independent AI observations to better understand how AI systems respond to a prompt.
Increasing the number of coordinated agents generally provides broader observation coverage, improves confidence in identified patterns, and reduces the influence of individual response variation.
Agent Count is therefore a measure of investigation depth, not simply computational scale.
Business Decision
Agent Count helps answer one strategic decision:
How much evidence do we need to make a confident decision?
Not every investigation requires the same level of observation.
Some business questions can be answered with relatively small investigations.
Others require broader evidence before meaningful conclusions can be drawn.
Business Questions
Agent Count helps organizations answer questions such as:
- How much evidence is sufficient for this investigation?
- How reliable are the observed patterns?
- Should this investigation prioritize speed or statistical confidence?
- When is a larger investigation justified?
- How should agent count be selected for different business objectives?
What Agent Count Represents
Each coordinated agent performs an independent observation within the investigation.
Together, these observations form the evidence used throughout Prompt Explorer.
Rather than attempting to identify a single "correct" AI response, Prompt Explorer analyzes patterns across many observations.
Increasing Agent Count generally allows the investigation to:
- Observe a broader range of AI behavior.
- Improve confidence in recurring patterns.
- Reduce the influence of isolated responses.
- Better represent probabilistic AI systems.
Agent Count should therefore be viewed as the depth of evidence supporting an investigation.
Why Agent Count Matters
Modern AI systems are probabilistic.
The same prompt may generate different responses depending on many factors.
Analyzing only one response may not accurately represent broader AI behavior.
By coordinating multiple observations, Prompt Explorer helps organizations distinguish between:
- Isolated responses.
- Meaningful behavioral patterns.
This provides a stronger foundation for investigation and decision making.
For more information:
Choosing an Appropriate Agent Count
The appropriate Agent Count depends on the purpose of the investigation.
Exploratory Investigations
When investigating a new topic or testing ideas, smaller investigations often provide sufficient directional insight.
These investigations emphasize speed and exploration.
Strategic Investigations
When investigating commercially important prompts, competitive positioning, or executive reporting, broader observation coverage is generally recommended.
These investigations prioritize stronger supporting evidence.
Ongoing Validation
Organizations frequently repeat investigations using consistent Agent Counts when validating optimization efforts or comparing historical observations.
Maintaining consistency improves comparability across investigations.
Balancing Speed and Confidence
Increasing Agent Count generally increases observation coverage.
It may also require additional processing resources and investigation time.
Organizations should select an Agent Count appropriate for the importance of the business decision being supported.
Prompt Explorer is designed to provide flexibility rather than prescribing a single investigation depth for every use case.
Best Practices
Match Investigation Depth to Business Importance
Not every prompt requires maximum observation depth.
Allocate larger investigations to prompts with greater strategic or commercial value.
Use Consistent Agent Counts for Comparisons
When comparing investigations over time, use the same Agent Count whenever practical.
Consistent observation depth makes historical comparisons more meaningful.
Interpret Results Alongside Evidence
Agent Count contributes to investigation reliability, but it should always be interpreted together with the supporting Evidence Layer.
Higher observation depth strengthens confidence, but does not guarantee identical AI behavior across future investigations.
Increase Depth Before Drawing Major Conclusions
If an investigation produces unexpected or strategically important findings, consider repeating the investigation with broader observation coverage before making significant business decisions.
Relationship to the Evidence Layer
Agent Count influences the amount of evidence available for an investigation.
The Evidence Layer explains how that evidence supports individual observations and findings.
Agent Count answers:
How much evidence was collected?
The Evidence Layer answers:
How strong is the evidence supporting this conclusion?
Together they provide the foundation for reliable prompt investigations.
Evidence Layer
Related Concepts
To better understand Agent Count:
For additional information about investigation reliability:
Related Pages
Agent Count works together with:
- Products → Prompt Explorer → Prompt Sets
- Products → Prompt Explorer → AI Models
- Products → Prompt Explorer → Evidence Layer
- Products → Prompt Explorer → Reading Results
These components collectively determine the quality and reliability of a Prompt Explorer investigation.
Next Steps
Once your investigation has been configured, continue with:
This guide explains how to interpret Prompt Explorer findings and transform observations into actionable investigation insights.