Aetherya

Thesia, inside your agent. The Aetherya MCP turns audience simulation into tools any agent can call.

Test a site, a headline or an idea against a synthetic audience without leaving the editor.

Read the docs
Capabilities

What your agentcan do with it.

Evidence from a synthetic audience, inside the loop, before a human reviews the work or a dollar is spent.

  1. 01Audit what it just builtYour agent audits a page, or walks a sign-up or checkout flow, and gets predicted friction, hesitation and drop-off back with concrete fixes.
  2. 02Decide before it buildsHand it a decision in plain English. Aetherya designs the study, builds the audience and quotes the price. Nothing runs until you approve it.
  3. 03Compare copy and creativeRank headline and ad variants, or pretest images and video, on predicted response, comprehension and trust, with the objections each one raises.
  4. 04Get a read in secondsCheck a handful of variants against a market population and see which archetypes favour each, before spending anything on the full study.
  5. 05Rehearse a postRun a post or an email through a synthetic audience first: predicted engagement, sentiment and representative reactions.
  6. 06Hand back a decisionAny finished run becomes a short brief with the answer, the reasoning, the risks and the next step, plus a link you can share.
Connect

Install via npm.Add your key.

The supported install path is the published @aetherya/mcp package, run on demand with npx, so it stays current.

Step 1 · Claude Code
terminal bash
claude mcp add aetherya --env AETHERYA_API_KEY=aeth_live_xxx -- npx -y @aetherya/mcp
Step 2 · Any MCP client
mcp.json json
{
  "mcpServers": {
    "aetherya": {
      "command": "npx",
      "args": ["-y", "@aetherya/mcp"],
      "env": { "AETHERYA_API_KEY": "aeth_live_xxx" }
    }
  }
}

Create a key in Settings → API keys and replace aeth_live_xxx. Usage is metered per simulation session against your plan. Results are simulated: every signal comes from a synthetic audience and predicted behavior, not real-user data.

Simulate audience cognition,from wherever you build.