Guides

Live Neon helps you answer three questions that most agent stacks blur together:

Who is this agent?
Why should I trust that identity?
How does that identity actually reach runtime?

These guides are for builders who want something more durable than prompt tweaking and more grounded than vibe-based personality design. If you are coming from hand-written system prompts, static personality files, or self-improving memory systems, the simplest way to understand Live Neon is this: prompts tell an agent what to do right now, memory stores what has happened, and Live Neon turns recurring evidence into a governable identity layer.

Live Neon dashboard showing agent identity overview with charts and metrics

Start here

Quickstart

The fastest path from blank workspace to a usable agent identity. Create an organization, define one agent, connect the right sources, run discovery, review what the platform found, and use the result in your runtime.

Core Concepts

A glossary and mental model for the platform. Read this if terms like Soul, belief, responsibility, signal, provenance, or dynamic prompt are still fuzzy.

Learn the model

PBD Pipeline

A deeper explanation of how raw material becomes structured identity. Covers distill, extract, compare, synthesize, and track, plus how it maps to observations, signals, beliefs, and provenance.

Content Sources

How to choose the right source mix for the agent you are building. Includes practical guidance for GitHub, websites, X, RSS, and LinkedIn, plus source hygiene and anti-patterns.

Put it to work

Integrations

How to use Live Neon with LangChain, CrewAI, Claude, OpenAI, coding agents, and custom runtimes. Includes patterns for static prompt generation, dynamic identity injection, and pairing identity learning with external systems.

Deployment

A production-minded guide for teams operating Live Neon seriously. Covers environment design, background jobs, scheduling, observability, and rollout strategy.

What these guides make obvious

A lot of the surrounding ecosystem focuses on one of three things:

Static personality templates - useful, but frozen until a human rewrites them
Failure memory - useful, but mostly about what went wrong
Ad hoc prompts - useful, but hard to inspect, govern, or scale across teams

Live Neon is strongest when you need a system for identity that can be learned, traced, reviewed, and shipped. That is why these guides keep coming back to the same ideas:

Identity should be earned through evidence
The system should keep full provenance
Recurring patterns should become beliefs
Templates can be useful starting points, but the best identity is the one the agent actually grows into

A useful mental model

Think of Live Neon as the layer that sits between experience and execution.

sources + conversations
        |
   observations
        |
     signals
        |
beliefs + responsibilities
        |
 resolved identity
        |
system prompt / dynamic prompt

The platform becomes much easier to reason about once you stop treating identity as a paragraph in a prompt and start treating it like a living system.

API Reference

Full REST API documentation with examples for every endpoint.