What is an Interlocute Node?
A node is a persistent, addressable unit of intelligence. It’s not locked to a single use case — it’s a flexible primitive you shape into whatever your product needs.
One primitive. Infinite forms.
Other platforms give you a chatbot, or an agent framework, or a pipeline tool. Interlocute gives you a node — and a node becomes whatever you tell it to be.
A Project
Like a ChatGPT project — a dedicated workspace with its own persona, memory, and conversation history.
An Agent
Independent, actionable, and autonomous. Give it tools, a persona, and a constitution — and let it act on your behalf.
A Chatbot
Embed it on your site, in a widget, via SMS, or WhatsApp. One node, many surfaces — same memory everywhere.
A Function Group
A logical grouping of capabilities. Combine RAG, tool use, guardrails, and scheduling into a single addressable endpoint.
A Media Processor
Feed it video, audio, images, or documents. It extracts, summarizes, and transforms media into structured intelligence.
A Scheduled Process
Set a cron. Your node wakes up, does its work, and goes back to sleep. Proactive intelligence without a server.
A Triggered Ingestion
React to incoming events — emails, webhooks, SMS — and process them through your node’s intelligence pipeline.
Something New
The node primitive is generic by design. Combine capabilities in ways we haven’t imagined yet — that’s the point.
Why generic?
Every AI platform makes a bet on what you’re building. Chat frameworks assume you want a chatbot. Agent frameworks assume you want tool-calling loops. Pipeline tools assume you want ETL.
Interlocute doesn’t assume. A node is the lowest-level primitive that still has identity, memory, and intelligence. You compose capabilities on top — and the node becomes exactly what your product requires.
The common stuff — conversation, scheduling, tool use, guardrails, streaming — is already built and production-hardened. You just turn it on.
Addressable
Every node has a stable identity. Call it by name from any surface, any system, any time.
Stateful
Threads, long-term memory, and context management are built in — not bolted on.
Configurable
Model, persona, capabilities, guardrails, scheduling — all live config. No redeploys.
Observable
Token counts, latency, cost attribution, and capability traces for every interaction.
Governed
IAM policies, constitutions, rate limits, and budget controls from day one.
The hard parts are done
Every node ships with a production-hardened foundation. You don’t build infrastructure — you configure behavior.
Conversation & Streaming
Multi-turn threads with SSE streaming, context windowing, and automatic state management. Works out of the box with any model.
Intelligence Capabilities
RAG, tool use, deep reasoning, memory, guardrails, and prompt templates. Toggle them on per node — no integration work required.
Surfaces & Triggers
API, web widget, SMS, email, WhatsApp, scheduled cron, webhooks. One node answers across every channel with shared context.
Think of it this way
A node is to intelligence what a web page is to information. It has an address. It has state. Anyone (or anything) with the right credentials can interact with it. And it can be anything.
Ready to build?
Create a node in seconds. Configure it for your use case. Call it from anywhere.