Data Activation
Platform & Process Overview

Working with CYTK
Data Requirements
Labeled & Structured
Your data should be organized relative to the main or top-level component (e.g. a vehicle, machine, component, etc.), and then further defined by labels (categories, positions, other qualifiers, data type, data format, etc.), which determine their structure.
Accessible
Ideally, data is accessible via a JSON Web API. We can discuss alternatives if needed.
API Evaluation and Data Assessment
Structure of the data is evaluated.
Ensure data can be scoped by model / component / product ID.
Ensure data can be searched and for each item of content, a unique identifier is present.
Proposal for New Product Development
Data Activation Pipeline

Key Platform Properties
ETL Service
Ingestion and processing of any structured and labeled data accessible via API or periodic downloads.
Automated, periodic Data Pipelines in charge of keeping specific data updated.
Structuring of data: Different sources of data are processed and refined to a common structure.
Loading into DBs and Elasticsearch
Machine Learning
Component Recognition
Industry-specific Lexicon
Data Integrity Checks
Extended M/L Roadmap
Related Searches
Technician Workflows
PDF Parsing
LLM Inclusion
Public Data Source
Generate structured results from OpenAI’s LLM as a separate data source to serve crowd-sourced web information where manufacturer service manuals may be underserving the technicans.
Search Results based on Proprietary Data training
Private RAG Pipeline to train OpenAI LLM on proprietary data in a private, secure setting to customize the technician’s experience relative to the specific data source.
Conversational Search (future)
Privately trained data sets would drive the LLM’s natural aptitude for real-time Q&A with technicians.