# Why Syntropic137?
> What makes Syntropic137 different from running agents ad-hoc in your terminal.

You can run an AI agent in your terminal right now. So why build a platform around it?

Because running agents is easy. Running them **repeatably, observably, and safely** — at scale — is hard. Syntropic137 closes that gap.

<FeatureGrid>
  <FeatureCard
    icon="workflow"
    title="Repeatable Workflows"
    description="Multi-phase pipelines built on the Claude Code command standard. Research, plan, implement, review — every workflow runs the same way, every time."
    gradient="indigo"
  />
  <FeatureCard
    icon="layers"
    title="Immutable Event Store"
    description="Every state change is a permanent, queryable event. Domain events, observability telemetry, and conversation logs — nothing is ever lost."
    gradient="purple"
  />
  <FeatureCard
    icon="git"
    title="GitHub-Native Triggers"
    description="Webhook triggers enable self-healing CI, auto-responses to review comments, and PR-driven workflows. Agents respond in minutes."
    gradient="cyan"
  />
  <FeatureCard
    icon="shield"
    title="Workspace Isolation"
    description="Every agent runs in an ephemeral Docker container. API keys are never exposed to workspaces. Egress proxies control outbound traffic."
    gradient="green"
  />
</FeatureGrid>

## The Problem

When you run Claude Code in your terminal, you get results — but you lose everything else:

- **No record** of what the agent did, which tools it called, or how much it cost
- **No way to replay** a successful workflow on a different repo
- **No isolation** — the agent has access to your entire machine
- **No triggers** — you have to manually kick off every task

## What Syntropic137 Adds

| Capability | Without Syntropic137 | With Syntropic137 |
|-----------|---------------------|-------------------|
| **Execution** | One-off terminal sessions | Multi-phase workflows with YAML definitions |
| **Observability** | Read the terminal output | Token usage, tool timelines, cost tracking, session replay |
| **Isolation** | Runs on your machine | Ephemeral Docker containers with secret injection |
| **Automation** | Manual start | GitHub webhooks, scheduled triggers, event-driven pipelines |
| **Data** | Gone when the terminal closes | Event-sourced — every state change is permanent and queryable |
| **Cost** | Check your API dashboard | Per-session, per-workflow, per-org cost breakdown |

## Who It's For

- **Engineering teams** using AI agents for code review, research, and automation
- **Platform engineers** who need observability and cost controls around agent usage
- **Solo developers** who want repeatable workflows without building infrastructure

## Next Steps

Ready to try it? [Get started in 5 minutes →](/docs/guide/getting-started)