Version: latest

Introduction to SENSA Fabric

This page describes what SENSA Fabric is, who uses it, and how it works.

What is SENSA Fabric?

The SENSA Fabric layer provides a collaborative platform for building, deploying, and managing trusted AI systems.

SENSA Fabric stack

Who should use SENSA Fabric?

Customers who want to accelerate in-house AI development: SENSA software helps clients automate and accelerate trusted AI development and deployment on any cloud.
Customers who want to manage and control AI systems: SENSA software helps clients protect their digital infrastructure and brand reputation from AI Business risks.
Customers who want AI powered predictions, processes, applications, and products: We use SENSA software and AI accelerators to build and deliver Industry AI solutions.

Platform support

SENSA Fabric can run on the following Kubernetes platforms:

AWS EKS
Azure AKS
Google GKE

Platform-specific details about installation and configuration can be found on the Cortex-charts docs site.

SENSA Fabric tools

CognitiveScale’s software tool-sets provide practical, scalable, trustworthy system for true end-to-end production-ready AI development.

SENSA Fabric Console - The heart of the SENSA system, includes:
- A visual orchestration interface where projects, profiles, campaigns, and agents are created, viewed, and managed
- A visual interface for managing connections, data-sources, secrets, and other components of profiles, campaigns, and agents
- Administrative pages where you can copy or download your personal authorization token, view your system-health dashboard, run an impact assessment, and perform other administrative functions
CLI - A command-line interface from which you can manage, monitor, and run SENSA Fabric.
VS Code: SENSA Fabric Developers Extension - A VS Code Extension that may be used to build and test Skills.
SENSA Python Libraries - The Cortex Python libraries offer Python control of administrative and development activities in SENSA Fabric ecosystem.
cortex-python - The base cortex-python library can perform most SENSA tasks, including assisting in data and model experimentation, setting up pipelines, and using Agents, Skills, and actions.
Profiles SDK - The SENSA Profiles SDK is collection of Java/Kotlin libraries, examples, and templates for utilizing SENSA Fabric to build Profiles and ingest profile data in a Spark-based environment, either on a local instance or in a SENSA Fabric Cluster.
CAMEL Specification - An open-source YAML variation that is used to configure Fabric components
SENSA REST API and GQL API - Use the APIs to directly manage, monitor, and run SENSA Fabric.

SENSA Fabric Building Blocks

Tecnotree's independently configurable, reusable building blocks provide real world AI for business solutions that can be easily configured.

Projects: The root organizational concept of the SENSA Fabric system. Profiles, Campaigns, Agents, Connections, and Secrets belong to Projects. You must select a project or set a project context to view, create, manage, and monitor Fabric components.
- Campaigns: A key organizational component of SENSA Fabric. Campaigns provide the framework for defining the goals, measurements, Data Sources, intervention, simulations, and plans that are executed and reported on by the system. (e.g. More patients participate in preventative medical services.)
  - Cohorts: The key actor of the campaign. Interventions and measures are set in place around the defined cohorts. Profiles and profile schemas are used to define cohorts. (e.g. Male patients over 50 years of age)
    - Cohort groups: Subdivisions of a cohort that can be assigned different missions and interventions. (e.g. Male patients over 50 years of age who have downloaded the medical provider app)
  - Goals: The measures that are configured for the mission or KPIs (Key Performance Indicators). The goals reflect how the business objectives for the Campaign will be met using data-driven objective values that can be observed. Each goal targets a specific cohort. (e.g. In 6 months there will be a 20% increase in prostate cancer screenings for the cohort)
    - KPIs: Objective measures of the goal that can be examined to determine the success of the mission interventions.
  - Missions: Missions are centered around a single goal state or desired outcome that relates to a selected goal-cohort and interventions configured to result in the goal state. Missions are used for both running simulations and deploying actions to production. Running a simulation creates a model that predicts/recommends plans (groups of interventions) for individuals in the cohort to reach the mission goal state. SMEs then review the plans, provide feedback on them, and re-simulate to generate a newer, better model.
    - Interventions: The specific activities composed of a cost estimate, pre-conditions, effects, and actions that are used to generate the models that are tested via simulations and deployed to production.
    - Plans: Groups of related interventions
    - Simulations: Application of the models to a subset of Cohort records that generates plans that SMEs review to provide retraining until the best intervention solution can be ascertained.
- Profiles: A second key organizational concept of SENSA Fabric. Profiles allows users to build, observe, and gain insights into entities by organizing attribute data that can be compared and leveraged by Campaigns and Agents.
  - Entities: A person, group, organization (or whatever) being profiled
  - Profiles: A consolidation of different pieces of information about a specific entity at different points in time
  - Attributes: The different pieces of information captured within profiles
  - Schemas: A set of attributes used to model a class of profiles.
  - Entity Events: Updates to the profile attributes and attribute values.
  - Versions: Every modification to a profile's attributes results in a new version of the profile. Versions help track how entities evolve over time.
- Agents: The third key organizational feature of SENSA Fabric. Agents provide automated AI-driven actions that are practical, scalable, and trusted.
  - Skills: The computational building blocks of an Agent. Skills are composed of actions (daemons or jobs) and models. Skill definitions identify input requirements and insight outputs. A typical ML (machine learning) model has two actions, one to train the model and one to provide insights. Skills are built and imported from any ML development environment using the cortex-python library.
  - Inputs/Outputs: Inputs can be Data Sources or payloads. Outputs can serve as inputs to other skills or they can deliver the payload of an agent or skill. Input-Output types are as follows
    - Services: Live links that send and receive data. Services are always defined in input/output pairs that require and deliver a payload to/from an agent.
  - Data flows: Agents are constructed by connecting Inputs, Skills, and Outputs. To provide for a seamless flow of data through the agent and make SENSA language and database agnostic, users map data types from Inputs to Skills and from Skills to Outputs using the data definitions provided in the CAMEL spec.

Connections: The databases or file-share systems used by Profiles, Agents, and Campaigns. SENSA Fabric v6 supports Mongo, Hive, S3, and files connection types.
Managed Content: SENSA Fabric managed content is used to store and retrieve data files for testing and training Agents and Skills, and for storing the output of Skills and Agents.
Container Registries: By default SENSA Fabric uses s3/Minio as its object storage backend. This is configured during DCI/cluster setup.
Data Sources: The backend data storage that Profiles pull data attributes and values from. Users are able to select and infer data features and develop schemas on-the-fly. Each Data Source pulls from a single Connection. A Connection may be referenced by more than one Data Source.
Secrets: Key-value pairs that are stored and called to provide access to data-sources or other resources outside of SENSA Fabric.