Usage journeys¶
Six expressions of intent-based computing
A personal system becomes useful through the life it supports
These journeys show how a person's intent can become a governed route, a native experience, and an outcome they can understand and control.
Every journey begins with a person's desired outcome. Unison resolves the relevant identity, context, policy, capabilities, and modality before composing an experience. Deterministic skills and tools provide reliable routes for known work. Governed inference helps pursue novel requests and can surface repeatable patterns for review as future capabilities.
Household operations¶
Intent: “Help me understand what needs attention at home this week and prepare the work I choose to do.”
- Governed context
- Household inventory, manuals, receipts, warranties, maintenance history, utility information, schedules, and approved sensor observations.
- Orchestration
- Product matching, maintenance rules, recall checks, cited retrieval, scheduling drafts, and governed inference for unfamiliar symptoms or equipment.
- Native experience
- A conversational briefing, visual workboard, Braille task structure, or another registered modality presents priorities, evidence, uncertainty, and available actions.
- Authority and safety
- The person approves sharing, scheduling, purchases, and physical actions. Device control requires a separate authorized actuation contract.
Unison can connect a photographed label to an owned appliance record, locate the exact manual, compare current symptoms with prior service, and prepare a cited brief. The person can correct the match, choose a safe next step, or ask for qualified help. The outcome includes the evidence used, decisions made, actions still awaiting confirmation, and a path to revise or cancel the plan.
Maturity: Being proven. Domain engines and synthetic workflows are implemented. Representative hardware, physical operation, and participatory evidence remain required.
Health, finance, and insurance¶
Intent: “Help me prepare for this care decision in the context of my coverage, budget, preferences, and health history.”
- Governed context
- Separately protected health, financial, insurance, and personal spaces contribute only the fields authorized for this purpose.
- Orchestration
- Source-preserving health timelines, benefit and cost rules, policy-document retrieval, range-based financial projections, and cited question preparation.
- Native experience
- The person receives a clear comparison of care options, likely coverage, affordability constraints, uncertainty, deadlines, and questions for clinicians or insurers.
- Authority and safety
- Clinical and financial professionals retain their roles. The person confirms every disclosure and consequential action, including provider messages, claims, payments, and appointments.
The health record, self-reported experience, insurer language, and inferred patterns remain visibly distinct. Unison can identify a coverage question, show the source clause, estimate a range from available information, and draft questions for the appropriate expert. Urgent-language rules prioritize timely human help. A purpose-bound receipt records which fields crossed context spaces, who authorized the use, and when that access expires.
Maturity: Being proven. Core domain and policy foundations have unit and simulation evidence. Live provider integrations, certification, and participatory evaluation remain future evidence gates.
Contextual research¶
Intent: “Research the options that fit my actual situation and help me decide what to do next.”
- Governed context
- The request can use approved goals, preferences, owned products, location, budget, prior decisions, and project records without exposing unrelated personal information.
- Orchestration
- Query planning, privacy-preserving external retrieval when permitted, local document search, source comparison, uncertainty tracking, and decision criteria selected by the person.
- Native experience
- Conversation, visual comparison, Braille navigation, or another modality conveys the same findings, evidence quality, tradeoffs, open questions, and next actions.
- Authority and safety
- External queries follow disclosure policy and use de-identification measures where available. Purchases, messages, bookings, and submissions remain explicit confirmation points.
Research remains connected to its sources and to the person’s stated criteria. Unison can preserve a useful investigation, revisit it when facts change, and explain why a recommendation changed. The result can become a cited brief, a decision record, or a proposed next action. Repeated research patterns can be proposed as reusable skills after review and testing.
Maturity: Implemented foundations with unit and hosted CI evidence. Provider-specific privacy controls and real-person usefulness require continuing evaluation.
Personalized education¶
Intent: “Help me understand this in a way that fits how I communicate, what I already know, and what I want to achieve.”
- Governed context
- Learning goals, demonstrated understanding, chosen curriculum, accessibility preferences, pace, interests, and guardian or educator policies where applicable.
- Orchestration
- Curriculum planning, retrieval from approved materials, adaptive explanation, practice generation, deterministic assessment rules, and governed inference for new questions.
- Native experience
- Concepts can become dialogue, tactile structure, visual demonstration, guided action, sign language through a future adapter, or another modality registered through the common contract.
- Authority and safety
- The learner controls pace, explanation style, and feedback. Applicable household, educator, and age-aware policies govern sources, sharing, and external actions.
Unison can carry an idea across sessions while keeping observed progress, self-reported confidence, and model inference distinct. A learner can ask for a different explanation, move between modalities, or explore a related question without losing the learning objective. The resulting record shows progress, sources, unresolved concepts, and the learner's chosen next step.
Maturity: Envisioned experience built on implemented orchestration and semantic foundations. Curriculum validation, representative modality testing, and participatory evidence remain required.
Shared understanding¶
Intent: “Help us understand the same situation and make a decision together through the modalities that work for each of us.”
- Governed context
- An explicit shared space contains the agreed evidence and task state. Each participant's private records, preferences, and modality state remain independently governed.
- Orchestration
- A common semantic outcome carries meaning, provenance, uncertainty, choices, confirmation, cancellation, and recovery across every expression.
- Native experience
- One person might use conversational and Braille interaction while another uses a visual and touch composition. Future sign language or BCI adapters connect through the same semantic contract.
- Authority and safety
- Each participant controls their own disclosures and confirmations. The shared action advances only with the authority required by its policy.
Equivalent control matters alongside equivalent meaning. Each person can inspect sources, understand uncertainty, pause the process, and participate directly. The shared receipt captures the agreed result while personal interaction data remains private. Explore the detailed shared understanding scenarios.
Maturity: Implemented semantic and modality contracts with unit and simulation evidence. Representative-device and participatory validation remains required.
Offline MacGyver mode¶
Intent: “The internet is unavailable. Help me understand this urgent household problem and find the safest useful next step with the resources I have.”
- Governed context
- Locally stored emergency references, household systems, manuals, shutoff locations, supplies, accessibility needs, and selected emergency information remain available during an outage.
- Orchestration
- A local model works with versioned offline knowledge packages, deterministic hazard rules, exact-device manuals, source provenance, freshness metadata, and jurisdiction-aware guidance.
- Native experience
- Concise conversation, visual steps, Braille instructions, alerts, or another available modality adapts to urgency, environmental constraints, and the person's interaction needs.
- Authority and safety
- Emergency services and qualified medical, gas, electrical, and repair help receive priority whenever available. High-risk guidance uses conservative safety controls and explicit confirmation.
The system can identify the relevant local shutoff record, present cited manual steps, explain uncertainty, and help prepare accurate information for a professional. It can also support routine repair and maintenance when the person chooses an appropriate route. Offline packages carry their source, version, review date, and update status so the person can judge their limits. The session receipt preserves what guidance was shown and which actions were confirmed.
Maturity: Envisioned safety-critical experience. Local inference and retrieval foundations exist, while curated knowledge packages, hazard evaluation, supported hardware evidence, and expert review remain required.
A common journey contract¶
These examples can grow as people discover new uses. Contributors can add a journey by preserving the same reviewable structure:
- State the person's intentDescribe the desired outcome, current situation, needs, preferences, and constraints.
- Identify governed contextName each context space, purpose, minimum disclosure, retention rule, and responsible authority.
- Compose the routeShow deterministic skills, tools, retrieval, models, inference bounds, and fallback behavior.
- Design native experiencesPreserve one semantic outcome while using the strengths of every supported modality.
- Define control and recoveryPlace confirmation, cancellation, receipts, incident handling, and professional escalation where they belong.
- Label the evidenceSeparate implemented behavior, work being proven, and envisioned capability using the evidence actually collected.
Evidence: These journeys combine implemented foundations, capabilities being proven, and envisioned experiences. Each journey states its maturity. The current status page is the source for repository-level evidence.