JamieOS records limited pseudonymous usage data to improve the service. It never records the content of your leadership conversations.
JamieOS Internal
JamieOS
This is the JamieOS internal API. It describes how the system processes inputs.
JamieOS routes every input through one of three operational modes: Clarity, Coaching, or Decision. The mode is passed as a query parameter (?mode=clarity|coaching|decision) when entering the chat screen. If no mode is provided, the system defaults to Clarity Mode. Each mode applies a distinct system persona and output structure: • Clarity Mode — Uses the Clarity Ladder (Facts → Meaning → Options → Decision → Action). Optimised for reducing noise and distilling priorities. • Coaching Mode — Applies the 5-Line Coaching Script and Behaviour Loop. Optimised for people leadership and 1:1 preparation. • Decision Mode — Uses trade-off analysis, risk framing, and decision matrices. Optimised for fast, confident decision-making.
Frameworks are structured thinking models that can be applied on demand within any chat session. They are surfaced via the Frameworks modal in the chat header. Available frameworks: • Clarity Ladder — Moves thinking from raw facts to committed action across five sequential steps. • Behaviour Loop — Diagnoses patterns using Trigger → Behaviour → Impact → Adjustment. • Leadership Lens — Examines situations through Intent → Behaviour → Outcome. • Reset Framework — A four-step protocol: Stop → Name → Reframe → Replace. Used for high-pressure moments. • 5-Line Coaching Script — Structures coaching conversations: Context → Observation → Impact → Question → Agreement. When a framework is selected, its associated prompt is injected into the chat input, allowing the user to customise and send.
JamieOS contains several standalone engines, each designed for a specific leadership task. Engines are accessible from the chat header or via dedicated screens. Available engines: • Clarity Engine — Distils complex or overwhelming input into a structured Clarity Ladder output. • Coaching Engine — Generates a structured coaching plan based on a described situation. • Decision Engine — Produces a trade-off analysis and recommended course of action. • Behaviour Diagnostic — Applies the Behaviour Loop to a described pattern or incident. • Reset Protocol — Runs the Stop → Name → Reframe → Replace framework on a stressful situation. • Team Diagnostic — Applies the Leadership Lens to assess team dynamics and leadership posture. • Interview Engine — Generates structured interview questions by category (Behavioural, Leadership, Role-Specific). • Content Engine — Produces formatted professional communications (emails, briefings, feedback, agendas). • Leadership Signals Engine — Generates decision-mode signals and risk indicators for an organisation. All engines invoke the LLM via the Core InvokeLLM integration with a mode-specific system prompt prepended.
Plans are structured records persisted to the Plan entity. They are created whenever a user selects "Save as Plan" from any engine, modal, or chat session. Plan schema: • title (string) — Auto-generated from the first 8 words of the AI response. • content (string) — Full AI-generated output. • created_at (string, ISO 8601) — Timestamp of when the plan was saved. Plans are accessible from: • Saved Plans screen (/saved) — Full list, sorted by creation date descending. • Dashboard screen (/dashboard) — Three most recent plans shown as a preview. Plans are user-scoped and listed using the base44 entities SDK. No deletion UI is currently implemented.
Every message sent and received in the chat screen is persisted to the Conversation entity. This enables session history, recent conversation views, and future context retrieval. Conversation schema: • session_id (string) — A unique identifier generated per chat session using timestamp + random string. • message (string) — The content of the message. • role (enum: user | assistant) — Who sent the message. • mode (string) — The active chat mode during the session. • timestamp (string, ISO 8601) — When the message was sent. Conversations are grouped by session_id for display. Sessions are surfaced on: • Recent Conversations screen (/recent) — Full session list with expandable message history. • Dashboard screen (/dashboard) — Three most recent sessions shown as a preview. Messages are written immediately upon send/receive and are not editable post-creation.