Build AI Agents That Actually Do Work

Stop writing boilerplate. Define what you want done, not how to do it. memra handles the AI orchestration so you can ship faster.

🚀 Quick Start

pip install memra

🔑 API Access Required: Contact info@memra.co for early access to the memra API

Next Steps:

Install the SDK: pip install memra
Get your API key from info@memra.co
Check out the examples below to see what you can build
Start with a simple agent and expand from there

🛠️ What Can You Build?

📄

Document Processing Pipeline

Auto-extract data from PDFs, invoices, contracts. Parse, validate, and push to your database.

Agent(role="Invoice Parser", tools=["PDFProcessor", "DatabaseWriter"])

📧

Customer Support Automation

Handle support tickets, categorize issues, draft responses, and escalate complex cases.

Agent(role="Support Analyst", tools=["EmailReader", "TicketClassifier"])

📊

Data Analysis Workflows

Connect to databases, run analysis, generate reports, and send insights to Slack.

                            Agent(role="Data Analyst", tools=["SQLQuery", "ChartGenerator", "SlackNotifier"])
                        

🔄

API Integration Chains

Chain multiple APIs together with AI decision-making between steps.

Agent(role="Integration Expert", tools=["HTTPClient", "JSONTransformer"])

🔍

Content Moderation Pipeline

Review user content, flag issues, apply policies, and maintain compliance automatically.

                            Agent(role="Content Reviewer", tools=["TextAnalyzer", "PolicyEngine", "FlagSystem"])
                        

🚀

Lead Qualification System

Score leads, enrich data from multiple sources, and route to the right sales team.

                            Agent(role="Lead Scorer", tools=["CRMConnector", "DataEnricher", "RouterAgent"])
                        

✨ Why memra?

📋

Declarative Design

Define workflows like Kubernetes YAML. Version control your AI business logic.

🤖

Multi-LLM Support

Seamlessly integrate OpenAI, Anthropic, and other providers with zero config changes.

💬

Self-Documenting

Agents explain their decisions in natural language for full transparency.

🏗️

Enterprise Ready

Battle-tested on real databases, files, and complex business workflows.

📡 API Endpoints

GET/health

Health check and API status verification

GET/tools/discover

Discover available workflow tools and their capabilities

POST/tools/execute

Execute workflow tools with structured input data

📚 Full Example: Invoice Processing

Here's how you'd build an invoice processing system that extracts data and updates your database:

from memra import Agent, Department
from memra.execution import ExecutionEngine

# 1. Define what you want done (not how)
invoice_processor = Agent(
    role="Invoice Processor",
    job="Extract vendor, amount, and line items from PDF invoices", 
    tools=[
        {"name": "PDFProcessor", "hosted_by": "memra"},
        {"name": "DataValidator", "hosted_by": "memra"},
        {"name": "DatabaseWriter", "hosted_by": "memra"}
    ]
)

# 2. Create a department (group related agents)
finance_dept = Department(
    name="Accounts Payable",
    agents=[invoice_processor]
)

# 3. Execute - memra handles the AI orchestration
engine = ExecutionEngine()
result = engine.execute_department(
    finance_dept, 
    {"file": "invoice.pdf", "database": "postgresql://..."}
)

# Result: Structured data extracted and saved to your DB
print(result.summary)  # "Extracted invoice #INV-001 for $1,234.56 from Acme Corp"

Ready to Build Your AI Workforce?

Join innovative teams automating their workflows with memra

Get Started Now