Dependencies & Environment
Relevant source files
The following files were used as context for generating this wiki page:
Introduction
The BrightPath AI Operations System is a lightweight, multi-agent infrastructure designed to automate inbound sales and customer service workflows. The environment is built primarily on Python and leverages the Anthropic Claude API for intelligent message processing, routing, and summarization.
The system's operational environment is comprised of a web-based dashboard for human oversight and a terminal-based orchestrator for direct processing. Data persistence is managed through a local SQLite database and a collection of Markdown-based memory files, ensuring that the agents maintain context across different sessions without the overhead of a heavy database management system.
Sources: README.md:1-12, app.py:12-25
Core Software Dependencies
The project relies on a minimal set of high-impact Python libraries to handle AI communication, user interface rendering, and terminal formatting.
Required Packages
The following dependencies must be installed via pip install -r requirements.txt:
| Dependency | Purpose | Version Requirement |
|---|---|---|
anthropic | SDK for interacting with Claude (Sonnet and Haiku models) | >=0.25.0 |
streamlit | Framework for the web-based Operations Dashboard | >=1.35.0 |
rich | Terminal formatting and enhanced logging for the orchestrator | >=13.0.0 |
python-telegram-bot | Optional dependency for manager briefing delivery | (Noted in README) |
Sources: requirements.txt:1-3, README.md:120-125, orchestrator.py:10-15
Environment Configuration
The system requires specific environment variables and directory structures to function correctly.
API Configuration
The primary requirement for running the pipeline is a valid Anthropic API Key. This can be provided via the environment or the UI:
- Environment Variable:
export ANTHROPIC_API_KEY=your_key_here - Dashboard Input: Users can enter the key directly into the Streamlit sidebar, which updates the global configuration state.
Sources: README.md:73-75, app.py:535-545
AI Model Selection
The system utilizes two distinct models from the Claude family to balance performance and cost-efficiency:
- Claude Sonnet (4.5): Used for specialist agents (Sales and Customer Service) where complex policy application and nuanced drafting are required.
- Claude Haiku (4.5-20251001): Used for the Triage and Manager Briefing agents to provide fast classification and summaries.
Sources: app.py:270-273, orchestrator.py:32-34
System Architecture & Data Flow
The environment is structured to support a linear agent pipeline. Each run follows a sequence from triage to specialist action, culminating in a manager briefing.
Pipeline Execution Flow
The following diagram illustrates how an inbound message traverses the environment's components:
The pipeline utilizes app.py for the web interface and orchestrator.py for terminal-based execution.
Sources: README.md:15-25, app.py:320-385, orchestrator.py:195-235
Data Persistence & Storage
The environment uses a hybrid storage approach to manage different types of operational data.
1. SQLite Database
A local database file (brightpath.db) is automatically created on the first run. It stores structured event logs and task tracking data.
| Table | Purpose |
|---|---|
events | Logs every pipeline run, including timestamps, categories, and raw agent outputs. |
tasks | Tracks generated actions with priority levels and statuses (open/closed). |
crm_log | Records every update made to customer records by the agents. |
Sources: app.py:134-155, orchestrator.py:46-65
2. Markdown Memory System
Agents read from and write to a set of .md files in the /memory directory. This provides a "human-readable" long-term memory.
company_context.md: Internal policies and pricing.customer_history.md: Per-customer interaction logs.sales_notes.md: Pipeline activity and lead status.open_tasks.md: Outstanding manual actions.
Sources: README.md:105-115, orchestrator.py:130-140
Development & Reset Tools
The environment includes utilities for maintaining the state of the system during demos or development cycles.
Reset Mechanism
The reset_demo.py script allows developers to clear the database and restore memory files to a baseline state using a backup folder.
Sources: reset_demo.py:15-45
Conclusion
The BrightPath environment is a portable, API-driven system that requires minimal setup beyond a Python 3.x environment and an Anthropic API key. By combining SQLite for structured logging and Markdown for agent memory, it provides a transparent and easily customizable workspace for AI-driven operations.
Sources: README.md:1-10, app.py:1-10