Privacy compliant chat interface for pdfs.
Reducing time-to-onboarding for consultants, PMs, TPMs and engineers into Cloud projects.
- Pdf document ingest (pdf size tbd)
- Multiple pdf document ingest (number of files tbd)
- Turn-based chat interface to converse against pdf content (memory size, number of turns, tbd)
- Hosted on scalable server (tbd which one)
- Explainability (tbd)
- More features (tbd)
Scenario 1: New engineer is onboarded to the Cloud team and needs to get up-to-speed quickly on a project. She collects all documents from Sharepoint and Github and converts them to pdfs. She uploads the pdfs to ChatPDF and is now able to ask key questions in a simple-to-use chat interface.
Scenario 2: New consultant ...
Scenario 3: New PM ...
Scenario 4: New TPM ...
The Content Similarity is for internal usage by the WTMAP only.
POC
- Python - programming language
- OpenAI - source of state-of-the-art LLMs
- Streamlit - web framework
- Lanchain - LLM framework
Deployment on scalable server
- Python - programming language
- OpenAI - source of state-of-the-art LLMs
- Streamlit - Web framework
- Google CloudRun - Google serverless service deployment platform
- Lanchain - LLM framework
- How do I ... (no one has asked yet)
All models and code is documented in the docs/ directory. insert link
It's important to be consistent with your naming conventions to ensure clarity and ease of navigation for developers and users accessing your documentation.
- Model Documentation:
- <model_name>_specification.md: Use the name of the model followed by "_specification" to indicate that it contains technical specifications for a specific model. For example, "image_classification_specification.md" or "sentiment_analysis_specification.md".
- Code Documentation:
- <code_topic>_documentation.md: Use a descriptive name that reflects the specific code topic or functionality. For example, "endpoint_deployment_documentation.md" or "data_preprocessing_documentation.md".