# Map a topic to Megido's niches before going further

> The niches closest to a topic, with their size, growth and labelling confidence, ready for the next step.

**Who asks:** Data engineer wiring Megido into a research agent

## Example

**Question:** Which Megido niches are closest to «staff shift scheduling for clinics»? Give me ids for the next steps.

**Megido calls:**

- `search_cohorts`: 3 niches

**Answer:**

```json
[
  {"niche_id": "<id>", "niche": "Medical and veterinary practice management SaaS", "level": 2, "size": "50-199", "share_growing": 36, "label_confidence": "high"},
  {"niche_id": "<id>", "niche": "Musculoskeletal healthcare providers", "level": 2, "size": "200-999", "share_growing": 33, "label_confidence": "high"},
  {"niche_id": "<id>", "niche": "Healthcare portals and medical centers", "level": 2, "size": "0-9", "share_growing": 33, "label_confidence": "low"}
]
```

None of them is about shift scheduling as such: the closest is clinic practice management software, which includes scheduling. The third niche has low labelling confidence; skip it.

Real data from the corpus, as of August 2026. Growth compares the latest month with the one before.

## How your agent answers

1. `search_cohorts` returned the three niches closest in meaning to the topic.

## Where it stops

These are the closest niches, not every fitting one. An empty or distant result means the topic is not covered, and the agent should say so. Ids are issued to each organization, so the example shows <id>.

## Ask your agent

```text
Which Megido niches are closest to «[topic]»? Give me ids for the next steps.
```