The Conversational State Transition Model
A neurocognitive model of how conversation changes what becomes thinkable and actionable
Read the paperTrust & Security
Privacy.
Precision.
Clear Boundaries.
Piloteer is built to improve team performance while protecting individual privacy, agency, security, and human judgment.
AccountablePeople make the decisions.
User controlPrivate guidanceEvidence before inferenceHuman judgment
Why we built it
Technology should make us more human.
Not less.
The promise of AI should not be to remove people from the equation. It should help us see each other more clearly, communicate better, make better decisions, and perform at a level that was harder to reach alone.
Piloteer was built around that belief. It gives people greater awareness in the moments that matter while keeping agency, privacy, and judgment where they belong: with people.
Trust by design
Private for the individual.
Aggregate for the organization.
Accountable at every decision.
01 / Private
Individual guidance stays with the user.
Piloteer is designed to give people private guidance and performance information that supports their own improvement.
02 / Aggregate
Leaders see team-level patterns, not individual scores or feedback.
Organizational visibility is designed to show the broader patterns affecting team performance without turning private individual guidance into a management scorecard.
03 / Accountable
Piloteer shows the evidence. People make the decisions.
Piloteer supports judgment with evidence, context, confidence, and uncertainty. Accountability remains with people.
Clear by design
Know when it’s active.
Know what it sees.
Know where the boundaries are.
01 / User control
Users start, pause, and stop Piloteer.
No hidden activation or background listening.
02 / Evidence before inference
Piloteer does not force conclusions from weak evidence.
Confidence, conflicting evidence, and what remains unknown stay visible. Low-evidence situations can result in no output rather than a forced interpretation.
03 / Human judgment
Piloteer provides evidence and guidance. It does not independently take consequential action.
People remain accountable for the decisions that follow.
Enterprise trust
Security, privacy, and governance are core to the system.
Security
- SOC 2 Type II
- Independent controls designed for enterprise assurance.
- Encryption
- Data is protected in transit and at rest.
- Enterprise SSO
- Support for enterprise identity and access management.
- Least-privilege access
- Access is limited to what is required.
- Audit logging
- Administrative and system activity can be tracked.
- Deployment options
- Secure AWS deployment by default, with customer-environment deployment available.
Privacy
- Customer data does not train shared models
- Customer data is not used to train Piloteer’s shared AI models.
- Purpose-bound data
- Piloteer processes the data needed to deliver the approved use of the system, not data simply because it is available.
- Focused on the Piloteer user
- Piloteer evaluates the person using it. Other participants may provide context to the interaction, but they are not individually evaluated or scored.
- Visual sensing stays on the device
- When visual sensing is enabled, camera imagery is processed on the user’s device and raw images or video are not retained.
- No stored audio or video
- Piloteer processes conversation in real time, but the Console does not create or retain raw audio or video recordings. Derived text and performance context can be retained to support guidance and improvement over time.
AI Governance
- Explainable, evidence-backed outputs
- Important observations should be connected to the evidence supporting them.
- Confidence + unknowns stay visible
- Piloteer preserves uncertainty rather than presenting every inference as fact.
- Bias + validation review
- Bias is treated as a measurable system risk, not a claim that any AI system is inherently bias-free. Piloteer uses structured validation, evidence thresholds, and human review as part of its control approach.
- Controlled change
- Models, measurement frameworks, and validation artifacts are versioned and reviewed before material changes reach production.
- Human oversight
- Piloteer augments human judgment rather than replacing accountability.
Direct answers
Common questions,
direct answers.
01Is Piloteer always listening?
No.
Piloteer operates when the user activates it. Users can start, pause, and stop the system, and there is no hidden background listening.
02Can managers see my individual Piloteer feedback or score?
No.
Individual guidance and scores stay private to the user. Leaders see aggregate team-level performance patterns rather than private individual feedback.
03Does Piloteer record meetings?
The Piloteer Console does not retain meeting audio or video recordings.
Audio and video may be processed to generate performance context, but the underlying Console meeting audio and video are not retained as recordings. If Piloteer Notetaker is used, it appears as a visible meeting participant and the recording behavior is clear.
04Does Piloteer use customer data to train its AI?
Customer data does not train Piloteer’s shared AI models.
Piloteer uses approved customer context to deliver the product without using customer data to train shared models.
05Does Piloteer analyze everyone in a meeting?
No.
Piloteer evaluates the Piloteer user. Other participants may provide context to the interaction, and how that context is represented can vary by product, but they are not individually evaluated or scored. When visual sensing is enabled, it applies only to the Piloteer user.
06Does Piloteer compare me to a universal behavioral profile?
No. Piloteer does not rely on a single universal behavioral profile.
Piloteer can use the individual’s own baseline and authorized historical context to make guidance more contextual and reduce dependence on one fixed population-wide norm. This personalization occurs through contextual inference rather than training a new model on the user.
07Does Piloteer make employment decisions?
No.
Piloteer is built to improve performance and support human judgment. It does not autonomously make hiring, promotion, compensation, disciplinary, termination, or other consequential employment decisions.
08How does Piloteer address bias?
Piloteer treats bias as a measurable system risk.
The system uses layered validation, evidence thresholds, contextual interpretation, controlled model changes, and human review rather than claiming that any AI system is inherently bias-free.
09What security controls does Piloteer support?
Piloteer supports enterprise security controls including SOC 2 Type II, encryption in transit and at rest, enterprise SSO, least-privilege access, audit logging, and enterprise deployment options.