emcie-co/parlant

Health v2 · maintenance only

Adoption not part of the score

18269 stars · 1554 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

Parlant is an open-source Python framework that acts as an interaction control harness for building reliable, customer-facing LLM agents. It dynamically assembles conversation context using primitives like observations, guidelines, tools, journeys, and canned responses to keep agent behavior consistent, compliant, and traceable at scale.

Use cases

• build a customer support chatbot that follows company policies
• deploy an LLM agent that won't hallucinate in customer conversations
• control what my AI agent says with granular behavioral rules
• replace brittle system prompts with structured agent guidelines
• build a compliant customer-facing AI agent for banking or healthcare
• attach backend API tools to an agent that only fire when relevant
• define multi-turn SOPs for an AI agent without rigid flowcharts
• trace and debug why my conversational agent said something

When to choose

• you need production-grade, predictable agent behavior for customer-facing conversations
• system prompts degrade as rules grow and you need structured context control
• compliance, brand voice, and traceability matter (B2C or sensitive B2B)
• you want to iterate on agent behavior from product feedback quickly
• you want an open-source alternative to Ada, Decagon, or Sierra

When to avoid

• you need simple internal chatbots where strict behavior control is unnecessary
• you want fully autonomous agents exploring tasks without conversational constraints
• your use case is non-conversational workflow automation rather than dialogue
• you need a lightweight prompt-chaining library rather than a full harness

Facets

framework · maturity active

agent-framework chatbot llm-inference rag prompt-engineering web-framework large-language-models chatbots developer-tools python self-hosted cross-platform conversational-ai context-engineering customer-service guardrails guidelines journeys behavior-as-code llm-orchestration compliance ai-agents retrieval-augmented-generation natural-language-processing docker

Member repositories

For agents

markdown · JSON · MCP: product_card(name="emcie-co/parlant")