Case study 1 of 3
DesignedCreate The Edge — AI Learning and Experience System
People and small organizations usually already hold the information needed to improve — it is just scattered across forms, conversations, documents, calendars, media, CRM records, and specialist knowledge. Learners receive generic content when they need a guided path, professionals repeat discovery and follow-up, and AI recommendations become unreliable when detached from consent, source material, and human judgment.
Design challenge
Four audiences, one system
People
- For the learner
- Make the next step understandable, personal and achievable without forcing the user to diagnose their own needs.
- For the professional
- Preserve review authority, professional scope and the ability to correct or override AI-supported recommendations.
Operations
- For the organization
- Turn fragmented information into repeatable onboarding, learning, communication and follow-up workflows.
- For the system
- Connect assessment, content, learner state, multimedia and operations while keeping permissions and actions auditable.
Core design principle. Begin with the person and the decision they need to make — not with a menu of disconnected tools.
Assessment and onboarding
Six-step flow
Assessment is treated as the beginning of a conversation, not an automated diagnosis.
- 1OrientationPurpose, expectations, AI disclosure, professional boundaries and consent.
- 2Structured assessmentGoals, context, barriers, preferences, confidence, constraints and relevant history.
- 3ClarificationThe AI Coach identifies gaps, asks focused follow-up questions and avoids unsupported conclusions.
- 4Human checkpointA professional reviews high-impact, ambiguous or scope-sensitive recommendations.
- 5Journey planThe learner receives a prioritized path with media, activities, check-ins and next actions.
- 6Adaptive reviewProgress, reflections and new information adjust the path without erasing prior decisions.
- Plain-language questions and progressive disclosure rather than one exhausting intake.
- Self-report, system inference and professional judgment stay separated in the learner record.
- Users control what is saved, shared or sent to a professional.
Personalized journey
Discover · Prioritize · Learn · Apply · Reflect · Adapt
Personalization inputs
- Learner context
- Goals, preferences, confidence, accessibility needs, timing and previously completed work.
- Knowledge context
- Approved content, source documents, learning objectives, assessment rules and professional guidance.
- Behavioral signals
- Completion, reflection, repeated questions, stated friction and requests for help.
- Human direction
- Professional review, learner choice, corrections, safety boundaries and escalation decisions.
Not a conversational bot
The coach is one component of a larger learning system: it reads learner state, follows an approved journey, retrieves governed content, creates structured next actions, and knows when to stop or involve a person.
- Interpret the current step using a structured profile and approved content.
- Recommend — not silently execute — high-impact actions.
- Identify the basis of substantive guidance where the interface supports it.
- Route ambiguity and professional-scope questions to human review.
Architecture
Experience, orchestration, services — governed throughout
Separating the visible experience from orchestration and services keeps a general-purpose model from becoming the system of record.
Learner experience
Assessment + consent
Goals, context, preferences, privacy choices
Personalized journey
Recommended path, media and activities
Progress + reflection
Check-ins, evidence, next-step choices
Orchestration
Learning rules
Sequencing and adaptation logic
AI coach / agent
Grounded guidance, retrieval and synthesis
Human review
Approval gates and professional escalation
Services + data
Backend data
Profile, progress, content and audit trail
CRM
Scheduling, messaging and follow-up
Media services
Avatar video and narrated audio
Governance across every layer. Consent · least-privilege access · privacy · content provenance · audit trail · human approval · safe escalation.
Interface and multimedia
Representative screen system
These describe the designed interaction model; they are not screenshots of a single deployed production build.
1. Onboarding
- Welcome + purpose
- Consent choices
- Assessment progress
- Save and continue
2. Learner home
- Current journey
- Today's recommendation
- Progress + reflection
- Message coach
3. AI coach
- Context-aware guidance
- Source-backed resources
- Action card
- Ask for human review
4. Review queue
- Flagged recommendation
- Learner context
- Approve / revise / escalate
- Audit note
Privacy and human-review controls
- Consent
- Explicit consent and AI disclosure at orientation.
- Access
- Least-privilege role and organization access; client-controlled sharing.
- Traceability
- Audit trail and content provenance on substantive guidance.
- Approval
- Gates for send, spend, contact and data exposure; escalation for clinical, legal, crisis or scope-sensitive content.
Current state — what is honest to claim
- Operational
- Public web experience: cohesive content, navigation, booking/assessment framing and client portals.
- Prototyped
- The command workspace: a working application with documented workflows; walk through after a current-state check.
- Designed
- Assessment, personalization, AI coach, governance and multi-tool orchestration shown in this case study.
- Planned
- Multi-tenant licensing, marketplace, production billing and broader integrations remain roadmap items.