INPUT01
Accessible input
Text, audio, documents and structured context enter through interfaces designed around the task and user.
AI(D)
A modular disability-aid platform that turns raw models and research into practical tools.
active development Modular effects chains · research-grounded tools · accessible interfaces
PB-06 active development
SYS/CHAIN 04 CHANNELS PALOMA BB
- Pronunciation
- Aye-Dee
- Model
- Modular chains
- Basis
- Evidence + research
- Role
- Practical support
01
02
03
04
What is real today
Useful assistance is assembled, not declared intelligent.
AI(D) is a modular disability-aid environment for turning raw models, research and small utilities into practical support tools. Its organizing metaphor comes from audio engineering: focused effects are connected into understandable chains for a particular person and task.
A chain may transform information, reduce friction, present choices or coordinate another tool, but each stage remains visible and replaceable. The interface is designed around access needs rather than exposing a generic model chat and asking the user to adapt.
The project is not a diagnostic authority and does not treat a model response as medical truth. Evidence, user control and clear boundaries matter more than performing intelligence.
System shape
Small capabilities become a purposeful support chain.
The environment separates input, transformation, evidence and presentation so each aid can be understood and adjusted.
CHAIN02
Modular effects
Each operation has a narrow purpose and can be reordered, replaced, bypassed or inspected.
EVID03
Research grounding
Tools can carry their evidence basis and limitations instead of presenting support choices as universal facts.
SURFACE04
Purpose-built interface
The final experience is shaped for the activity rather than exposing implementation details as the product.
Current evidence
The user remains the authority over the aid.
AI(D) is structured to make support adaptable without allowing a model or preset to silently become the decision-maker.
01
Composition
Chain-based
Capabilities are assembled from explicit stages rather than buried in one general assistant.
02
Adjustment
Per-stage
Individual transformations can be tuned or disabled without discarding the complete workflow.
03
Interface
Task-specific
The experience can be built around the aid rather than around a chat transcript.
04
Authority
User-held
Suggestions remain support, not diagnosis, coercion or hidden automated judgment.
Operating principles
Constraints that keep the work honest.
Support, not authority
The system assists decisions without claiming ownership of them.
Visible transformations
Every stage of a workflow can be understood and changed.
Access before spectacle
Interfaces are shaped around actual friction and need.
Evidence has boundaries
Research grounding includes uncertainty and scope.
Connected work