Agent Substrate
Features

Everything you need.
Nothing you don't.

A complete stack for building, deploying, and monitoring AI agents. No glue code required.

Agent Engine

Production-ready Python framework for autonomous AI agents. Built on FastAPI with full async support, PostgreSQL persistence, and Redis-backed memory.

  • ReAct agent loop (Think → Act → Observe)
  • Multi-agent orchestration & planning
  • 15+ built-in tools (web, code, email, files, etc.)
  • Guardrails: PII detection, content filtering, prompt injection
  • Memory: unbounded, sliding-window, token-limited
  • Structured output parsing with Pydantic
  • LLM-as-judge evaluation framework
  • Python 3.13, fully async, type-safe
ReAct Agent Loop
iteratedoneUser PromptInput GuardThinkActObserveWeb SurferCode ExecTask MgrOutput GuardResponse
chat
Extract all invoices from the uploaded folder
Thinking...I'll process each PDF. Found 12 invoices. Extracting fields...
extract_invoice
⏳ Awaiting approval...

Chat Platform

Deploy conversational AI interfaces with real-time streaming, interactive tool widgets, and human-in-the-loop approvals.

  • Real-time SSE streaming with token-by-token display
  • Human-in-the-loop tool approvals
  • Interactive MCP App widgets via iframes
  • Kanban task board with live updates
  • Voice: speech-to-text, TTS, real-time S2S
  • File upload/download with drag & drop
  • Thread management with full history
  • Dark/light theme with responsive design

Visual Builder

Design agent workflows as directed graphs. Drag nodes onto the canvas, configure parameters, connect edges, and test in real-time.

13 Node Types

Everything you need to compose complex agent workflows.

A

Agent

LLM reasoning step with configurable model & prompt

T

Tool

Tool invocation with risk level and HITL mode

S

Skill

Reusable prompt-skill packages

G

Guardrail

Input/output validation layer

R

Router

Classify & branch based on conditions

M

Memory

Session memory backend (Redis/Postgres)

M

MCP

External MCP server connection

C

Condition

Expression-based branching

A

Approval

Human-in-the-loop gate

W

While

Loop construct with max iterations

S

Start / End

Flow anchor points

N

Note

Annotation for documentation

document-processor.pipeline
urgentnormalStartGPT-4o AgentSchema GuardRouterEmail SenderTask ManagerEnd
Catalog

15+ built-in tools.

Ready-to-use tools for common agent tasks. Extend with your own or connect MCP servers.

Web Surfer

Browse and extract web content

Code Interpreter

Sandboxed Python in Firecracker

Doc Analyzer

Parse PDFs, invoices, receipts

Ask Human

Pause and request human input

Task Manager

Create & manage Kanban boards

Email Sender

Send transactional emails

Knowledge Search

RAG vector search

Pipeline Runner

Execute sub-pipelines

Observability

Every agent run is traced end-to-end. See exactly what happened, when, and why. Structured JSON logs, distributed traces, and pre-built Grafana dashboards.

  • OpenTelemetry tracing → Grafana Tempo
  • Structured JSON logging → Loki
  • Prometheus metrics for latency, throughput, errors
  • Pre-built dashboards: Services Overview + Frontend Logs
  • Per-request trace IDs across all services
observability.log
INFOagent_runtimeAgent "analyst" started run r-3f8a
INFOtool_executorExecuting web_surfer (iteration 1/10)
WARNguardrailsPII detected in output, redacting...
INFOagent_runtimeRun r-3f8a completed (3 iterations, 2.4s)
trace_id: a1b2c3d4e5f6
batch-job.json
Batch Job: Invoice Extraction87% complete
invoice_001.pdf$4,250.00
invoice_002.pdf$1,890.50
invoice_003.pdf...
invoice_004.pdf—

Batch Processing

Submit bulk jobs, monitor progress in real-time, and export results. Perfect for document extraction, data analysis, and content generation at scale.

  • Submit CSV/JSON batches via API or dashboard
  • Real-time progress tracking per item
  • Automatic retry on failures
  • Export results as CSV, JSON, or via webhook
  • Rate limiting and concurrency controls