WebGithub

Integrations

DigitalFate is designed to work seamlessly with the tools, services, and platforms you already use. Whether you're embedding it into SaaS applications, scaling on the cloud, or extending it with custom tools, integration is at the heart of the DigitalFate architecture. It's built to adapt, extend, and connect not to lock you in.

⚙️ Supported LLM Providers

DigitalFate provides out-of-the-box compatibility with multiple foundational models and providers:

Provider
Models Supported
Integration Details

OpenAI

GPT-4o, GPT-4, GPT-3.5

OPENAI_API_KEY

Anthropic

Claude 3.5 Sonnet, Claude 2.1

ANTHROPIC_API_KEY

DeepSeek

DeepSeek Chat

DEEPSEEK_API_KEY

Azure OpenAI

GPT series with Azure endpoints

AZURE_API_KEY, endpoint, version

AWS Bedrock

Claude 3.5 via Amazon Bedrock

AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_REGION

pythonCopyEditclient.set_config("OPENAI_API_KEY", "your-key")
client.default_llm_model = "openai/gpt-4o"

You can switch models per task or globally per client instance.

☁️ Cloud Platform Integration

DigitalFate is cloud-ready and works on:

  • AWS: Deploy agents or entire MCP clusters with Bedrock and Lambda support

  • GCP: Use container-based workflows and run inference via Vertex AI

  • Azure: Integrate with OpenAI models via Microsoft’s secure endpoints

Production-ready deployments are available using:

  • Docker

  • Kubernetes

  • Serverless functions

bashCopyEditdocker run -p 8000:8000 digitalfate/server

📚 Tool Ecosystem

Tools are modular Python-based wrappers that allow agents to perform real-world tasks:

  • Included Tools:

    • Search: Web search

    • Browser: Simulate UI interactions

    • PDFReader: Extract content from PDFs

    • CSVTool: Read structured data

    • CodeInterpreter: Run Python code in a sandbox

    • ReviewTool: Evaluate task outputs

  • Add Custom Tools:

pythonCopyEditfrom digitalfate.client.tools import Tool

class MyCustomTool(Tool):
    def run(self, input_text):
        return custom_logic(input_text)

task.tools = [MyCustomTool]

Custom tools integrate seamlessly with task execution and agent reasoning.

🔐 Secure API Integrations

DigitalFate supports API-level authentication and secure key management for external services. Tools can call internal APIs, webhook-based services, or third-party APIs with encrypted credentials.

  • OAuth support (planned)

  • Environment variable key loading

  • Encrypted config file support

📥 Knowledge Base Sources

You can inject context from:

  • PDF files

  • URLs (public or gated)

  • Web scraping (via Browser)

  • Company documents (via API or upload)

  • Internal tools (via custom adapters)

pythonCopyEditfrom digitalfate import KnowledgeBase

kb = KnowledgeBase(
    files=["whitepaper.pdf"],
    urls=["https://openai.com/research"],
)
agent.knowledge_base = kb

This allows agents to reason using custom company data or proprietary research.

🧠 Memory & Long-Term Context (Advanced)

DigitalFate’s memory system can be connected to:

  • Redis (in-memory persistence)

  • Vector DBs (future feature)

  • File-based or encrypted local caches

This allows agents to remember:

  • Past tasks

  • User preferences

  • Ongoing goals

Memory is scoped per agent and configurable for secure, production use.

🔁 SaaS & API Integration

You can embed DigitalFate into your app or stack using:

  • RESTful endpoints (via client.call() and server APIs)

  • Webhooks

  • Scheduled tasks / CRON

  • Task queues (RabbitMQ, Celery, etc. integration-ready)

Examples:

  • Trigger an agent workflow via form submission

  • Schedule reports via agent execution

  • Automate ticket classification via LLM response

🔄 CI/CD and DevOps Friendly

Use DigitalFate inside your pipelines:

  • GitHub Actions

  • GitLab CI

  • Jenkins jobs

  • Airflow DAGs

Perfect for:

  • Testing AI pipelines

  • Auto-generating docs from source

  • ML evaluation reports

🔧 Developer-Friendly SDK

The SDK is written in Python and structured for quick extension. Every major function (agents, tools, tasks, servers) can be subclassed and overridden.

Supports:

  • Async execution

  • Server-client split

  • Logging & tracing hooks

  • Custom agent reasoning loops

🧪 Future Integrations (Coming Soon)

  • Zapier + Make workflows

  • Slack / Discord bots

  • LangChain-compatible tools

  • RAG pipelines with Pinecone & Weaviate

  • Frontend embedding with React + Next.js

  • OpenAPI spec-based tool auto-importing

PreviousCollaboratingNextSocial links

Last updated