v0.1 — Private build · Australia

Vision intelligence for the venues that move the city.

MeridianAI builds edge hardware and computer vision systems that turn the floor of a nightlife venue into structured, real-time data. Capacity, dwell, queue, and flow — measured in milliseconds, never stored as footage.

~12ms
Edge inference latency
0
Identifiable footage stored
24/7
Floor-state telemetry
2026
First cohort, NSW

Venues run on instinct. Instinct misses things.

Capacity is guessed at the door. Bar staff are scheduled to a number nobody measured. Marketing spend hits a venue with no idea whether the floor was full or empty when the campaign ran. The data already exists — it just hasn't been captured.

/ 01

Operators flying blind

Door clickers, manual headcounts, and end-of-night POS reports tell you what already happened. By then, the decision window is closed.

/ 02

Marketing without ground truth

Promoters and agencies optimise content with no feedback loop into actual venue occupancy. Campaigns get judged on impressions, not floor lift.

/ 03

CCTV isn't analytics

Existing camera systems were built for evidence, not insight. Cloud video pipelines are slow, expensive, and create privacy liability nobody wants.

One device. Every floor signal.

A small purpose-built edge unit mounts in-venue, processes vision locally, and streams structured telemetry to a single dashboard your operators, promoters, and agency partners all read from.

  • Real-time capacity

    Live headcount per zone, per door, per room. Updates faster than your bar staff can pour.

  • Dwell & flow

    How long people stay, where they cluster, which rooms drain when. Floor-shape data nobody has had before.

  • Queue intelligence

    Live queue length and movement at each entry point — feed it to door staff, ticket apps, or a queue-skip integration.

  • Privacy by architecture

    All inference runs on-device. The cloud receives counts and vectors — never identifiable footage. PII never leaves the room.

// floor_state live
412
people on floor · 78% capacity
22:47 fri node_03 · main_room

From the rafters to the dashboard in milliseconds.

Three steps. No NVRs to rip out. No tape on the lens.

01 / INSTALL

Mount the node.

One Meridian edge unit per zone, powered over ethernet. Installed in an evening — no integration with your existing CCTV stack required.

02 / INFER

Process on-device.

Custom vision models run locally on the node. Frames are transformed into anonymous structured data — counts, vectors, dwell — before anything ever leaves the venue.

03 / ACT

Read it everywhere.

One API powers your operations dashboard, your queue-skip app, your marketing partners' reporting, and your weekly investor PDF. Single source of truth.

Built for the way nightlife actually runs.

A platform shaped by operators, not retrofitted from retail analytics.

Operations

Capacity dashboards

Live floor state across every zone. Push thresholds to floor staff via Slack, Discord, or a tablet at the door.

Marketing

Campaign attribution

Tie content posts and promoter pushes to actual occupancy lift in the hours after they ran. Finally answer "did it work?"

Door

Queue-skip integration

Native pipe into queue management apps and ticketing systems. Live queue data drives dynamic pricing and door flow.

Staffing

Predictive scheduling

Six weeks of floor data is enough to forecast next Saturday. Rosters that match what the night actually does.

Reporting

Owner reports

Weekly PDFs that read like a brief, not a spreadsheet. Built for operators who want signal, not dashboards.

Platform

Open API

Every signal Meridian captures is available via REST and webhooks. Build on top of it, or let us pipe it where you need.

What we won't compromise on.

Building infrastructure for venues means earning trust before measuring anything.

  • Privacy is a hardware decision.

    Vision processing runs on the node. The cloud sees counts and vectors — never people. No facial recognition, no re-identification, no exceptions.

  • Latency is a product feature.

    If the data arrives after the moment passed, it's a liability, not a tool. Every signal is engineered to fire under 50ms end-to-end.

  • Operators own the data.

    Your floor data is yours. Export it, pipe it, walk away with it. Open formats, open API, no lock-in clauses.

  • One number wins.

    We'd rather ship one signal a venue actually acts on than ten dashboards that go unread. Calibrated outputs over noisy ones.

Common questions, answered straight.

What does MeridianAI do?
MeridianAI builds edge hardware and computer vision systems that give nightlife venues real-time intelligence on capacity, dwell time, queue length, and floor flow. Vision is processed on-device — no cloud video, no identifiable footage stored.
How is this different from CCTV analytics?
CCTV pipelines were built for evidence and run video to the cloud. MeridianAI nodes process vision locally and only emit structured numerical signals. That means lower latency, lower cost, and a privacy posture you can actually defend at a council hearing.
What kind of venues is this built for?
High-volume nightlife: pubs, clubs, late-night bars, and multi-room venues where capacity, queue management, and peak-hour decisions move real revenue. Australian venues are the first cohort.
Where is MeridianAI based?
MeridianAI is based in Australia, with initial deployments across NSW. We're operator-first by design — co-built with venues, not pitched to them.
When can I install it?
MeridianAI is in private development. Venues can request early access through the form below. The first cohort begins late 2026 — small, hands-on, with founder support throughout.
How much does it cost?
Pricing is being shaped with the first cohort. The model will be hardware-included, monthly subscription per node — sized to make sense for an independent venue, not just a chain. Specifics on the call.

Be in the first ten venues that get one.

Spots are limited and matched personally. If your venue runs Friday and Saturday nights at scale, we want to talk.

No spam. One reply, from a founder.