GATHER FOR DEVELOPERS

Your customer.
As an API.

Every customer-facing decision, validated before it ships. Gather answers a question about your buyers with a verdict and a confidence score your code can branch on, in minutes instead of weeks. Ship when the answer is strong. Escalate to real interviews when it is not.

Get an API key Connect over MCP

REST and MCP. Bearer auth. JSON in, findings out.

A check your pipeline can read, not a report someone has to.

before-you-ship.sh

# Ask your buyers before the campaign goes live
curl https://synthetic.gatherhq.com/api/v1/studies \
  -H "Authorization: Bearer $GATHER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "topic": "Does leading with onboarding speed beat
              leading with price for this segment?",
    "study_type": "messaging",
    "audience_description": "Directors+ in IT and DevOps at
       North American companies with 500-5,000 employees"
  }'

# -> headline_finding, key_findings[], consensus_themes[],
#    objections[], report_url - in one response

WHERE IT GOES IN YOUR WORKFLOW

A check that runs before you ship.

You would not ship code without a test. Every one of these decisions is normally made on a guess and measured weeks later. Put a call in front of it and the guess becomes an answer while you can still act on it.

01
Before ads go out
Score creative and hooks against the audience you are actually buying, while the budget is still movable.
02
Before emails send
Test the subject line and the offer on the segment the campaign targets, not on last quarter’s list.
03
Before the page ships
Put two positioning options in front of buyers and ship the one they can repeat back to you.
04
Before pricing changes
Find the objection and the walk-away point before the change reaches a renewal conversation.

TWO WAYS IN

Call it from your code, or hand it to your agent.

The REST API is for pipelines and product code. MCP is for Claude, Cursor and any agent that should be able to ask your customers a question on its own.

REST API

One POST, one verdict.

Bearer auth against synthetic.gatherhq.com. Every study comes back with overall_confidence, so the call is a gate your code can act on rather than a document to read.

gather.ts

const res = await fetch(
  "https://synthetic.gatherhq.com/api/v1/studies", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.GATHER_API_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    topic: "Which launch angle lands with CISOs?",
    study_type: "messaging",
    audience_description: "CISOs and VP Security at" +
      " North American enterprises, 1,000+ employees",
    webhook_url: "https://yourapp.com/hooks/gather",
  }),
});

const study = await res.json();
console.log(study.headline_finding);
console.log(study.key_findings);
console.log(study.report_url);

MCP

Your agent asks the customer.

Gather is a connector in the Claude directory. Point any MCP client at it and your agent can run real research, pull findings and cite a report without leaving the chat.

claude_desktop_config.json

{
  "mcpServers": {
    "gather": {
      "command": "npx",
      "args": [
        "-y", "mcp-remote",
        "https://share.gatherhq.com/api/mcp"
      ]
    }
  }
}

// Then just ask, in plain language:
//
//   "Before we launch Tuesday's campaign, run a
//    messaging study on mid-market IT buyers and
//    tell me which angle to lead with."
//
// Sign in with your work email on first connect.
// The agent runs the research and comes back with
// findings, a confidence score and a report link.

GET STARTED

Copy this. You are running in a minute.

Same call, your language. Set GATHER_API_KEY and go. The Node example shows the pattern that matters: branch on the confidence score instead of reading the answer.

bash

# 1. See what you can ask
curl https://synthetic.gatherhq.com/api/v1/study-types \
  -H "Authorization: Bearer $GATHER_API_KEY"

# 2. Run a study against your own audience
curl -X POST https://synthetic.gatherhq.com/api/v1/studies \
  -H "Authorization: Bearer $GATHER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "topic": "Will a 15% increase change renewal intent?",
    "study_type": "pricing",
    "audience_description": "Directors+ in IT, DevOps and
      Infrastructure at North American companies with
      500-5,000 employees who use observability tools"
  }'

# 3. Browse everything you have run
curl https://synthetic.gatherhq.com/api/v1/studies \
  -H "Authorization: Bearer $GATHER_API_KEY"

REFERENCE

The whole surface area.

Base URL https://synthetic.gatherhq.com. Every route takes Authorization: Bearer YOUR_API_KEY.

POST/api/v1/studiesRun a study. Preset audience or a natural-language audience_description. Returns findings, themes, objections and a report URL.
GET/api/v1/studiesList completed studies with headline findings and report links.
GET/api/v1/studies/{id}Fetch one study in full, including every interview response.
POST/api/v1/researchLighter call for agents: topic plus audience, returns a markdown summary ready to paste.
POST/api/v1/simulateRun a scenario against your brand’s own saved personas in Gather.
GET/api/v1/personasEvery preset B2B and B2C profile, with the default sets.
GET/api/v1/study-typesThe eleven study types: brand, product, concept, competitor, cx, pricing, messaging, winloss, churn, segmentation, custom.
MCPshare.gatherhq.com/api/mcpThe Gather connector, listed in the Claude directory. Work-email sign-in on first connect, then one conversational tool that reaches everything Gather can do.

FOUR THINGS TO BUILD FIRST

Put it in the path of a decision.

AD PRE-FLIGHT
Nothing launches unasked
In the campaign build step, run a messaging study on the audience you are about to buy. Hold the launch if the angle does not land.
creative ready → POST /studies → objections → launch or rewrite
EMAIL PRE-SEND
Subject lines that earn the send
Score two or three subject lines against the exact segment the send targets, then pass the winner to your ESP.
segment picked → POST /studies → winner → ESP
PRICING
Find the walk-away first
Run a pricing study before the change reaches a renewal call. Get the objection and the point where intent breaks.
proposed price → POST /studies → walk-away point → decide
AGENTS
Give your agent a customer
Connect over MCP and your agent stops guessing what buyers want. It asks them, then writes the brief itself.
agent task → run_research → findings → draft

START BUILDING

Your customers, one call away.

Tell us what you are building and we will issue a key and walk the first integration with you. Most teams have a study running the same day.

Get an API key Talk to our team

Your customer.As an API.

A check that runs before you ship.

Before ads go out

Before emails send

Before the page ships

Before pricing changes

Call it from your code, or hand it to your agent.

One POST, one verdict.

Your agent asks the customer.

Copy this. You are running in a minute.

The whole surface area.

Put it in the path of a decision.

Nothing launches unasked

Subject lines that earn the send

Find the walk-away first

Give your agent a customer

Your customers, one call away.

Your customer.
As an API.