Agentic Readiness

As AI browser agents (like AutoGPT, Chrome AI Agents, and web automation subagents) become standard web clients, the nature of web form interaction is evolving.

AI agents fill complex multi-page applications, compose long support responses, and execute workflows on behalf of human users.

---

🤖 Why Agents Need ContinueJS

AI agents face the exact same disruptions that human users experience:

1. Context Window / Session Timeouts: An AI agent executing a multi-step job application may pause while waiting for human authorization or API tokens.
2. Tab Crashes & Rate Limits: If an automated browser session crashes or is interrupted by network rate limits, all programmatic input is lost unless stored locally.
3. Multi-Step Workflows: Agents often navigate across multiple wizard steps where state needs to persist across steps without requiring immediate database commits.

---

⚡ Agent-Agnostic API Design

ContinueJS makes zero assumptions about whether the entity typing into input fields is a human user or an AI agent.

// An AI agent filling out a job application programmatically:
const draft = createDraft({ id: 'agent-application-step-1' });

// Agent sets form values programmatically
draft.set('fullName', 'John Doe');
draft.set('yearsOfExperience', 5);
draft.set('coverLetter', 'Generated summary...');

// Agent saves draft before proceeding to the next step
await draft.save();

// If the agent process restarts or resumes later:
const savedProgress = await draft.restore();
// Agent resumes seamlessly!

Key Principles for AI Agent Compatibility

• Programmatic Input Interception: FormObserver tracks synthetic input and change events generated by AI agents.
• Framework Agnostic: Works regardless of whether your frontend is built with React, Vue, Svelte, or Vanilla Web Components.
• JSON Serialization: Draft state is serialized in clean JSON key-value pairs that LLM context windows can easily consume and summarize.