Responsible AI and Contributor Data

Welcome to Positron Flux

Positron Flux is a software-delivery excellence workbench. It brings together metadata from the tools involved in software delivery, measures delivery outcomes, identifies relationships between engineering practices and those outcomes, and helps teams choose practical improvements.

Some connected records contain contributor names. Positron Flux handles that information deliberately: individual activity is presented as factual activity, while automated analysis, diagnostics, and recommendations focus on teams and delivery systems. The product does not convert engineering telemetry into an overall judgement of a person.

What to expect

Positron Flux is designed to provide useful evidence without creating a workplace-surveillance or automated people-management system.

Customers and contributors can expect:

Defined metrics. Each metric has a specific meaning, scope, source, and time period.
Faithful reporting. Contributor-level information describes events recorded by connected systems; it does not claim to measure invisible work or personal qualities.
Team-level analysis. Models identify patterns in team practices, delivery conditions, and outcomes.
Actionable improvement. Recommendations concern changes to workflows, collaboration, ownership, tooling, and delivery practices.
Human oversight. Positron Flux provides evidence and recommendations; it does not make employment decisions.
Responsible AI. Product behaviour is bounded to prevent individual performance assessment, behavioural profiling, emotional inference, and employment recommendations.
Respect for the EU AI Act. Positron Flux is designed and operated in accordance with the Act’s requirements and prohibitions relevant to AI in the workplace.^1,2

These boundaries are part of the product’s design and behaviour. They do not depend solely on each customer remembering to apply an acceptable-use policy after deployment.

Contributor information

Connected source-control, delivery, and operational tools naturally record who participated in an event. Positron Flux may display source-based facts such as:

Recorded deployments, builds, releases, commits, and pull requests
Review, approval, assignment, and participation events
Active branches older than a customer-defined threshold
Completion of a clearly defined engineering outcome
Badges based on explicit and visible rules
Leaderboards for a specific recorded activity, scope, and period

This information retains its narrow, factual meaning. It is not presented as an assessment of a contributor’s productivity, performance, capability, effort, quality, value, or suitability for an employment decision.

For example, “most deployments this month” means only that the connected systems recorded the highest number of deployment events for that contributor within the selected scope and period. It does not mean “most productive engineer” or “best performer.” Activity is one dimension of engineering work and cannot serve as a complete measure of developer productivity.³

Badges and leaderboards follow the same rule. They may recognise or order explicit, observable facts; they do not rank contributors by inferred overall performance.

Team-level intelligence

Positron Flux uses automated analysis to understand software-delivery systems. Its models can identify relationships between engineering practices and selected delivery outcomes, compare team-level results with relevant benchmarks, highlight risks, and recommend practical changes.

Analysis may consider:

Delivery cadence and flow
Work in progress and cycle time
Review, queue, and handoff delays
Build and release reliability
Incidents, rework, and operational load
Dependencies and delivery bottlenecks
Knowledge concentration and continuity risk
Team-level conditions associated with sustained workload pressure

Typical product outputs may explain why a team is above or below a benchmark, compare teams within an appropriate context, identify practices associated with a desired outcome, suggest changes that could improve that outcome, or highlight team-level continuity and sustainability risks.

The unit of interpretation is the team and its delivery system. Recommendations therefore address matters such as workflow design, work-in-progress limits, review practices, documentation, shared ownership, pairing, rotation, training, staffing conditions, and recovery capacity.

A knowledge-concentration signal, for example, may indicate an opportunity to strengthen documentation or distribute ownership. It is not a judgement that a particular contributor is indispensable, overvalued, undervalued, or safe or unsafe to dismiss.

A team sustainability signal may indicate conditions such as excessive operational load, frequent interruption, persistent work in progress, inadequate recovery time, or staffing pressure. It is not a diagnosis of any contributor’s physical, emotional, or mental state.

A whole-team perspective

Positron Flux applies the principle behind the SPACE framework: developer productivity is multidimensional and cannot be understood from activity counts alone.^4,3

SPACE dimension	Positron Flux perspective
Satisfaction and wellbeing	Team-level conditions affecting whether work appears sustainable, without inferring an individual’s emotions or mental state
Performance	The value, quality, reliability, and outcomes of the team’s work
Activity	Recorded engineering events used as contextual evidence, not as a complete measure of a contributor
Communication and collaboration	Reviews, coordination, knowledge sharing, handoffs, and dependencies
Efficiency and flow	Delays, interruptions, bottlenecks, work in progress, and how effectively work moves through the delivery system

This approach produces a more useful account of software delivery while avoiding simplistic conclusions about individuals.

Product boundaries

Positron Flux does not generate or present:

Individual productivity, performance, capability, value, or employability scores
Rankings based on inferred overall individual performance
Claims that a named contributor is slowing down or accelerating a team
Behavioural or personality profiles of contributors
Assessments or predictions of an individual’s burnout, wellbeing, mood, engagement, or emotional state
Inferences about emotion from code, text, voice, facial expressions, activity patterns, or other signals
Recommendations about who should be hired, promoted, rewarded, disciplined, reassigned, or dismissed
Conclusions that a contributor is safe or unsafe to dismiss because of knowledge concentration or “bus factor”
Automated allocation of work based on inferred individual behaviour, personality, or personal characteristics

These are intentional safeguards, not omitted analytics. They apply across dashboards, reports, alerts, recommendations, badges, leaderboards, model outputs, and any other way Positron Flux presents information.

EU AI Act

Positron Flux respects the EU AI Act. Its AI capabilities are designed for team-level software-delivery improvement, not for the evaluation or management of individual workers.

The EU AI Act identifies certain employment and workers-management systems as high-risk, including AI intended to monitor or evaluate workers’ performance or behaviour, allocate tasks based on individual behaviour or personal characteristics, or make decisions affecting work relationships, promotion, or termination. Positron Flux does not provide AI capabilities for those purposes.²

The Act also prohibits AI used to infer a person’s emotions in the workplace, apart from limited medical or safety exceptions. Positron Flux does not perform workplace emotion recognition and does not infer a contributor’s emotional state from engineering activity or other data.¹

Positron Flux reflects these requirements in concrete product behaviour:

Contributor-level records remain descriptive and source-based.
Automated assessments, correlations, diagnostics, and recommendations operate at team or delivery-system level.
Activity volume is never presented as a complete measure of an individual’s productivity or value.
Workload and sustainability signals describe team conditions rather than individual health or emotions.
Knowledge-concentration signals lead to resilience recommendations rather than personnel recommendations.
Product outputs do not support individual employment decisions.
Automated insights remain subject to human review and organisational context.

The EU AI Act also requires providers and deployers to support an appropriate level of AI literacy among staff and others who operate AI systems on their behalf. This onboarding guidance is part of Positron Flux’s commitment to making the product’s capabilities, limits, and intended use clear.⁵

Using insights responsibly

Positron Flux insights are most effective when they prompt questions about the environment in which teams work:

Where is work waiting, being interrupted, or repeatedly handed off?
Which practices are associated with the outcome the team wants to improve?
How can ownership and knowledge be shared more safely?
What changes could make delivery more reliable, effective, and sustainable?
What operational context should be considered before acting on a signal?

Contributor activity counts should not be turned into personal targets. Doing so would remove the context that gives the data meaning and would conflict with the intended use of Positron Flux.

Our commitment

Positron Flux is built to improve software delivery while respecting the people doing the work. Contributors are represented accurately and narrowly; managers receive evidence about teams and delivery systems; and organisations retain human responsibility for decisions affecting people.

Positron Flux helps customers understand the work, improve the system, and protect the distinction between engineering evidence and judgement of an individual.