Real-world grounding for AI products.

InBetween gives LLMs, copilots, and agents a structured, verified layer of hospitality data: queryable attributes, preference matching, and group logic designed for function calling.

Get Started

User request

Natural-language ask lands in the agent.

Restaurant for six after an event — veggie, near the station.

Restaurant for six after an event?
Veggie options please
Within 10 min of the station
1.4s

Thinking & tooling

Resolve constraints, then call InBetween.

Thinking…
Parse party size & dietary
Geocode station proximity
Next: call venue search toolagent

The Lantern Room

Match found · 4 min walk

Veggie
Quiet
Tables for 6
Near station

The Lantern Room — quiet, veggie-friendly, ££, 4 min walk, tables for six.

Grounded reply to user0.2s

For AI founders, product leaders, and engineering teams

Models reason well. Give them facts to reason over.

When someone asks an AI product where to eat, the model's fluency is not the constraint. The constraint is data: without structured, current venue attributes, the most articulate answer in the world is still a guess.

InBetween is that data layer for food and drink. Dietary coverage, atmosphere, noise, seating, and multi-person travel fairness are exposed as typed parameters a model can call directly.

The schemas were designed for function calling from the start, so integrating InBetween is closer to declaring a tool than building an integration.

What breaks today

  • Ungrounded models produce confident, wrong venue recommendations.
  • Generalist POI data lacks the attributes consumer recommendations need.
  • In-house enrichment pipelines are a permanent tax on the engineering team.

What you can ship

  • One hospitality grounding layer across every AI surface you ship.
  • Restaurant and bar recommendations your product can stand behind.
  • Location-aware features shipped in days rather than quarters.

How it works for you

  1. 01

    Declare InBetween as a tool

    Typed parameter schemas drop into any function-calling framework.

  2. 02

    Let models query real attributes

    Dietary compatibility, atmosphere, distance, and group constraints resolve against verified data.

  3. 03

    Return verifiable answers

    Shortlists are grounded in maintained records, not generated text.

api.inbe.us/v1/search200 OK
POST
{
"locations": [{ "lat": 51.51, "lon": -0.12 }]
"mode": { "mode": "POINT", "maxDistance": 800 }
"preferences": {
"catersFilters": ["vegetarian"]
"goodFor": ["groups", "cocktails", "conversation"]
}
}
response
"establishments": [{
"displayName": "The Lantern Room"
"distanceMeters": 320
"catersFlags": ["vegetarian", "vegan"]
"matchScore": { "total": 0.94 }
}]
Venue assistant

Restaurant for six after an event?

Veggie options please

Within 10 min of the station

The Lantern Room — quiet, veggie-friendly, ££, 4 min walk, tables for six.

The Lantern Room

Match found

Veggie✓
Quiet✓
4 min walk✓
Tables for 6✓

Every venue. Every detail.

Precision metadata for apps that help people meet, eat and drink.

Venue intelligence

Ultra detailed venue data

Get detailed information about venues, including menus, pricing, ambiance, and more.
Smart matching

Preference driven

Wow your customers with recommendations and information tailored for them.
Multi-location search

Tools for multiple locations. Find restaurants, bars and cafés based on multiple sets of preferences and locations.

In-app experience

Loyal to you

Keep users inside your app by giving them the info they need to make decisions.
Decision support

Built for decision-making

A suite of tools designed to enable you to answer customer needs directly.

Grounding for AI

Give your models facts worth reasoning over.

Open up a new world for your users with InBe's preference-aware venue discovery—built for group venue matching, dietary search, and ranked shortlists.

No card required. Full core API access.

Start building for freeExplore the API

Keep exploring

Questions teams ask

Built for developers who connect people to places

Start buildingExplore the API

Connecting users to real-world places shouldn’t mean stitching geocoding, routing, Places search, and ranking yourself. One preference-aware search replaces the pipeline.

api.inbe.us/v1/search200 OK
POST
{
"locations": [
{ "lat": 51.523, "lon": -0.158, "label": "Alex" },
{ "lat": 51.493, "lon": -0.098, "label": "Sam" }
],
"mode": { "mode": "POINT", "maxDistance": 750 },
"preferences": {
"goodFor": ["date", "cocktails", "conversation"],
"catersFilters": ["vegetarian"],
"establishmentFilters": ["servesCocktails"]
}
}
response
{
"centerPoint": { "Lat": 51.508, "Lon": -0.128 },
"radius": 750,
"establishments": [
{
"displayName": "The Lantern Room",
"distanceMeters": 142,
"matchScore": { "total": 0.94 }
},
{
"displayName": "Barrio Soho",
"distanceMeters": 186,
"matchScore": { "total": 0.91 }
}
]
}

Fair middlegrounds

Multi-person locations in one call; we balance travel so you don’t build routing yourself.

Preferences that actually match

Dietary, vibe, good-for, and constraints as first-class inputs, not post-filters on pins.

Pay for the hard parts

Core search and enriched place data stay cheap; you pay where decision intelligence earns its keep.

Tools for user decisions

Ranked shortlists and experience-ready answers so your product keeps users in-flow, not off to a map.

Illustrative scenario

0ms

Ranked shortlist in one call

Preference-aware search returns a decision-ready shortlist without stacking Places, routing, and enrichment calls.

~15× faster vs multi-call pipelinesNot a live benchmark
RESTWorks with any stack
curl -X POST https://api.inbe.us/v1/search \
  -H "API-Key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "locations": [
      { "lat": 51.523, "lon": -0.158, "label": "Alex" },
      { "lat": 51.493, "lon": -0.098, "label": "Sam" }
    ],
    "mode": { "mode": "POINT", "maxDistance": 750 },
    "preferences": {
      "goodFor": ["date", "cocktails", "conversation"],
      "catersFilters": ["vegetarian"],
      "establishmentFilters": ["servesCocktails"]
    }
  }'
Full examples in docs

Start in minutes with the knolage you can scale without stress.

Generous free usage to evaluate, startup credits to grow, and a migration path when you are ready to switch from stacked Places pipelines.

Start in minutes

No sales call. Sign in, grab a key, and ship a first search.

Start building

Build — free forever

£10 of usage every month, full core API, no card required.

See Build plan

InBe for Startups

£2,500 usage credit, Grow-level features, and onboarding support.

Startup programme

Switch with confidence

Migration credits from £500, dual-running help, and high-spend matched offers.

Migration programme