Is it worth pursuing?
What client outcome matters? Where could AI extend human judgement, and what must remain human-owned?
Illustrative demonstrationFictional organisation · synthetic data · composite employee voices
PEOPLEPOTENTIAL
CONVERSATIONAL INTELLIGENCE · AI–HI USE CASE
People Potential / Illustrative evidence hub
Explore the idea.
Shape how it could work.
Before committing to AI-assisted client work, explore the proposal with the people who would make it work. Their experience helps reveal what to protect, what to change and what to test before implementation.
People Potential interpretation · Conversations powered by Haven, a FeedbackRocket innovation
START WITH A POSSIBILITY
Explore an idea with the people who would make it work.
Ask how the proposal fits the organisation before investing in delivery. Early involvement can build familiarity and readiness while surfacing practical requirements and concerns.
What would enable it?
Which decision rights, quality standards, resources and management practices would the pilot require?
How would it land?
Where would it help daily work? What exceptions, extra load or risks do the people doing the work anticipate?
In this demonstration, the findings describe the starting conditions. They guide design and trials; they do not demonstrate successful implementation.
ONE ENGINE · FOUR LAYERS
A question becomes evidence for a decision.
01
Detection
Client + People PotentialFrames the proposed use case and selects tensions and layers to explore. Directional signals target the enquiry; they are not a diagnosis.
02
Extraction
Haven · a FeedbackRocket innovationUses adaptive follow-ups to explore experience, examples, concerns and conditions for the idea to work.
03
Identification
People Potential · human interpretationPeople Potential interprets patterns and affect in business context, forming hypotheses about process, people and organisational layer.
04
Recommendation
People PotentialTurns working interpretations into sequenced options with authority boundaries, measures and assumptions to test.
WHY THESE OUTPUTS ARE DISTINCTIVE
More than a theme list or a sentiment score.
Traceable interpretation
Connect each finding to anonymised evidence and its strategic, tactical or operational layer. Separate observed experience from a possible explanation.
How it is experienced
The Affective Signal Codebook distinguishes 11 feelings and 6 language registers, with cited words, intensity and uncertainty. Pride, guardedness and resignation can call for different responses to the same process issue.
A practical route to test
The Intervention Mapping Playbook adds a response family, sequence, delivery adjustments, moves to avoid and re-measurement markers. People Potential tailors these to the organisation.
Evidence and reporting discipline
Participants review and approve their conversation summary. Reporting uses agreed anonymisation and cohort boundaries. Codes describe speech, not a person. Thin coverage and scripted positivity lower confidence; ambiguity stays visible. Conversations support hypotheses and do not establish cause.
Method basis: AI–HI Integration Brief; Affective Signal Codebook v1.0; Intervention Mapping Playbook v1.0. The numbers and voices shown here are synthetic, not coded client findings.
THE TENSION ARCHITECTURE
What would it take to make this idea work here?
The same illustrative findings now inform a proposed pilot. Select a tension to connect the signal, organisational layer, participant evidence and possible response. These are working hypotheses, not established causes.
Live — priority for design and testing
Watch — emerging; keep under review
Monitored — present, but not a current priority
The number is a friction signal out of 100: the nearer it is to 100, the stronger or more concentrated the pressure. It is not a percentage of employees.
Live tension
Human–AI relationship
Human Judgment vs Automation
63/100 friction signal
Human Judgment
Automation
Where does accountability end and the algorithm begin?
Strategy66
Tactics69
Operations55
LIKELY ORIGINStrategy & Tactics
WORKING INTERPRETATIONBoth: process and people
WHAT MEASUREMENT SHOWS
Informal AI use is already present, but authority boundaries, disclosure expectations and accountability are not consistently understood.
WHAT CONVERSATION REVEALS
People are using judgment, but often privately. Unclear rules are producing both concealment and unnecessary checking.
Perception gap
22% of participants are unsure who carries accountability when an AI-supported recommendation is wrong.
“I am accountable for the recommendation, but nobody has told me when I am expected to disagree with it.”Illustrative composite employee voice
“I disclose AI use to people I trust. I do not know whether the firm wants transparency or plausible deniability.”Illustrative composite employee voice
DIRECTION TO TEST
Name the authority model for each use case, establish a safe challenge protocol and make disclosure expectations explicit.
Monitored
Human–AI relationship
Trust vs Dependency
52/100 friction signal
Trust
Dependency
When does reliance begin to erode the judgment it should support?
Strategy40
Tactics52
Operations47
LIKELY ORIGINBelow prioritisation threshold
WORKING INTERPRETATIONMonitored
WHAT MEASUREMENT SHOWS
The signal is present, but it is not strong or concentrated enough to justify a full intervention in this demonstration.
WHAT CONVERSATION REVEALS
Conversation adds context without changing the prioritisation decision. The tension remains visible for the next pulse.
Monitored
Human–AI relationship
Legibility vs Judgment
49/100 friction signal
Legibility
Judgment
What is lost when tacit expertise is forced into explicit criteria?
Strategy37
Tactics49
Operations44
LIKELY ORIGINBelow prioritisation threshold
WORKING INTERPRETATIONMonitored
WHAT MEASUREMENT SHOWS
The signal is present, but it is not strong or concentrated enough to justify a full intervention in this demonstration.
WHAT CONVERSATION REVEALS
Conversation adds context without changing the prioritisation decision. The tension remains visible for the next pulse.
Watch signal
Human–AI relationship
Capability vs Readiness
58/100 friction signal
Capability
Readiness
Are tools moving faster than practice, culture and identity?
Strategy46
Tactics58
Operations53
LIKELY ORIGINUnder observation
WORKING INTERPRETATIONWatch signal
WHAT MEASUREMENT SHOWS
The signal is present, but it is not strong or concentrated enough to justify a full intervention in this demonstration.
WHAT CONVERSATION REVEALS
Conversation adds context without changing the prioritisation decision. The tension remains visible for the next pulse.
Monitored
Organisational design
Agency vs Control
51/100 friction signal
Agency
Control
Do people retain genuine authorship over AI-assisted work?
Strategy39
Tactics51
Operations46
LIKELY ORIGINBelow prioritisation threshold
WORKING INTERPRETATIONMonitored
WHAT MEASUREMENT SHOWS
The signal is present, but it is not strong or concentrated enough to justify a full intervention in this demonstration.
WHAT CONVERSATION REVEALS
Conversation adds context without changing the prioritisation decision. The tension remains visible for the next pulse.
Live tension
Organisational design
Centralisation vs Empowerment
72/100 friction signal
Centralisation
Empowerment
Where do decision rights actually sit - structurally, not on paper?
Strategy38
Tactics78
Operations71
LIKELY ORIGINTactics
WORKING INTERPRETATIONProcess problem
WHAT MEASUREMENT SHOWS
Decision authority is understood at executive level, but becomes inconsistent when translated through management.
WHAT CONVERSATION REVEALS
Employees are not resisting accountability. They are learning decision rights through rejected attempts and repeated escalation.
“The policy says I own the outcome, but every unusual decision still travels two levels up.”Illustrative composite employee voice
“You learn where your authority ends by having the work sent back.”Illustrative composite employee voice
DIRECTION TO TEST
Design explicit decision rights, then introduce AI-suggested routing for low-risk approvals while managers retain authority.
Watch signal
Organisational design
Accountability vs Psychological Safety
56/100 friction signal
Accountability
Psychological Safety
Can people disclose AI errors without losing accountability?
Strategy44
Tactics56
Operations51
LIKELY ORIGINUnder observation
WORKING INTERPRETATIONWatch signal
WHAT MEASUREMENT SHOWS
The signal is present, but it is not strong or concentrated enough to justify a full intervention in this demonstration.
WHAT CONVERSATION REVEALS
Conversation adds context without changing the prioritisation decision. The tension remains visible for the next pulse.
Monitored
Organisational design
Individual Autonomy vs Collective Coherence
47/100 friction signal
Individual Autonomy
Collective Coherence
When does local adaptation become organisational fragmentation?
Strategy35
Tactics47
Operations42
LIKELY ORIGINBelow prioritisation threshold
WORKING INTERPRETATIONMonitored
WHAT MEASUREMENT SHOWS
The signal is present, but it is not strong or concentrated enough to justify a full intervention in this demonstration.
WHAT CONVERSATION REVEALS
Conversation adds context without changing the prioritisation decision. The tension remains visible for the next pulse.
Monitored
Complexity & sensemaking
Certainty Seeking vs Uncertainty Navigation
54/100 friction signal
Certainty Seeking
Uncertainty Navigation
Are tools for certainty being forced onto genuinely complex work?
Strategy42
Tactics54
Operations49
LIKELY ORIGINBelow prioritisation threshold
WORKING INTERPRETATIONMonitored
WHAT MEASUREMENT SHOWS
The signal is present, but it is not strong or concentrated enough to justify a full intervention in this demonstration.
WHAT CONVERSATION REVEALS
Conversation adds context without changing the prioritisation decision. The tension remains visible for the next pulse.
Watch signal
Complexity & sensemaking
Shared Meaning vs Multiple Perspectives
59/100 friction signal
Shared Meaning
Multiple Perspectives
Do teams interpret AI outputs consistently enough to act together?
Strategy47
Tactics59
Operations54
LIKELY ORIGINUnder observation
WORKING INTERPRETATIONWatch signal
WHAT MEASUREMENT SHOWS
The signal is present, but it is not strong or concentrated enough to justify a full intervention in this demonstration.
WHAT CONVERSATION REVEALS
Conversation adds context without changing the prioritisation decision. The tension remains visible for the next pulse.
Monitored
Complexity & sensemaking
Efficiency vs Resilience
45/100 friction signal
Efficiency
Resilience
Is optimisation stripping out the redundancy that protects delivery?
Strategy33
Tactics45
Operations40
LIKELY ORIGINBelow prioritisation threshold
WORKING INTERPRETATIONMonitored
WHAT MEASUREMENT SHOWS
The signal is present, but it is not strong or concentrated enough to justify a full intervention in this demonstration.
WHAT CONVERSATION REVEALS
Conversation adds context without changing the prioritisation decision. The tension remains visible for the next pulse.
Live tension
Execution rhythm
Speed vs Quality
68/100 friction signal
Speed
Quality
Is the deadline conversation overriding the quality commitment?
Strategy51
Tactics73
Operations79
LIKELY ORIGINStrategy–Tactics boundary
WORKING INTERPRETATIONPeople problem
WHAT MEASUREMENT SHOWS
Quality remains the stated priority, while delivery behaviour, escalation and recognition consistently reward speed.
WHAT CONVERSATION REVEALS
The strategy is not missing. Its transmission is. Teams experience the deadline conversation as the real instruction.
Perception gap
31% of coded passages within this tension describe the stated quality commitment as out of step with lived delivery pressure.
“Leadership says quality comes first, but every conversation about a missed deadline tells us speed is what matters.”Illustrative composite employee voice
“In our informal AI use, the first draft is faster. We still have three reviews and the rework after them.”Illustrative composite employee voice
DIRECTION TO TEST
Keep the intervention human-led: reset management signals, clarify quality thresholds and measure whether rework and the perception gap move.
Watch signal
Execution rhythm
Standardisation vs Flexibility
57/100 friction signal
Standardisation
Flexibility
Where does consistency begin to remove necessary human nuance?
Strategy45
Tactics57
Operations52
LIKELY ORIGINUnder observation
WORKING INTERPRETATIONWatch signal
WHAT MEASUREMENT SHOWS
The signal is present, but it is not strong or concentrated enough to justify a full intervention in this demonstration.
WHAT CONVERSATION REVEALS
Conversation adds context without changing the prioritisation decision. The tension remains visible for the next pulse.
Monitored
Execution rhythm
Innovation vs Operational Stability
50/100 friction signal
Innovation
Operational Stability
Is AI disrupting workflows that were already working?
Strategy38
Tactics50
Operations45
LIKELY ORIGINBelow prioritisation threshold
WORKING INTERPRETATIONMonitored
WHAT MEASUREMENT SHOWS
The signal is present, but it is not strong or concentrated enough to justify a full intervention in this demonstration.
WHAT CONVERSATION REVEALS
Conversation adds context without changing the prioritisation decision. The tension remains visible for the next pulse.
Monitored
Execution rhythm
Transparency vs Information Overload
55/100 friction signal
Transparency
Information Overload
Is AI generating more usable signal - or more noise?
Strategy43
Tactics55
Operations50
LIKELY ORIGINBelow prioritisation threshold
WORKING INTERPRETATIONMonitored
WHAT MEASUREMENT SHOWS
The signal is present, but it is not strong or concentrated enough to justify a full intervention in this demonstration.
WHAT CONVERSATION REVEALS
Conversation adds context without changing the prioritisation decision. The tension remains visible for the next pulse.
What “perception gap” means here
Within a tension, it is the fraction of coded passages where an employee describes the organisation’s stated position as out of step with lived experience. It is not the percentage of employees who disagree.
FROM FINDINGS TO IMPLEMENTATION CHOICES
Use the hypotheses to design the next move.
The illustrative findings suggest clarifying decision rights and management signals before extending AI use. Leadership can accept, adapt, investigate or decline each option. Implementation is a separate engagement.
ONE EVIDENCE-TO-ACTION EXAMPLE
Decision delays may be a design issue.
Evidence: people learn authority boundaries through rejected requests. Hypothesis: unclear decision rights at the Tactics layer contribute to delay.
Test: co-design a named ownership map and trial one low-risk approval flow, with AI suggesting and a manager authorising. If resignation is evident, deliver a small visible improvement before introducing the wider redesign; avoid a large programme announcement with no follow-through.
Read the outcome: compare cycle time, escalation and quality with the baseline; use Conversational Intelligence Pulse to explore clarity, trust and whether the change is felt. Revise the hypothesis if the evidence does not support it.
CHOOSE · TRIAL · LEARN
Build an implementation plan from what you learn.
Agree a bounded trial
Define the scope, ownership, resources and measures, then put the agreed change into practice. Agree human authority, the baseline, success criteria and conditions for pausing before work begins.
Listen alongside delivery
Conversational Intelligence Pulse →
Powered by Haven, a FeedbackRocket innovation.
Explore how people experience the trial and what needs attention. Test a focused assumption before it begins, listen during delivery or revisit a finding afterwards. Read this alongside operational evidence.
Reframe what comes next
Bring the evidence together, reconsider assumptions and decide what to adapt, investigate or pursue next. Compare outcomes with the hypothesis and report back what was heard, what changes and why.
These are connected options, with each engagement scoped to what you need.
What would we watch in this example?
Approval time and escalation alongside decision clarity; rework and quality exceptions alongside the stated-versus-experienced priority gap. Protect professional pride and client commitment as work changes.
More candid frustration can sometimes reflect returning engagement rather than deterioration. Examine the language and context across the cohort, alongside delivery evidence, before judging progress. No post-implementation results are claimed here.
BEGIN WITH YOUR BUSINESS QUESTION
Make the next decision better informed.
Explore a proposed use case, understand the conditions it would need and shape a practical route into action.