Case study

Lumeo

How Lumeo ships 1,000 deploys a day with 50+ autonomous agents on Hives.

Explore the details

From a frame to an event you can use.

Scrub an illustrative video-analysis timeline. The synthetic scene explains the workflow; it contains no real footage and makes no claim about Lumeo's measured results.

Video event inspectorSynthetic scene

object_01

Example zone A

Frame event00:01

Frame received

The example frame enters the analysis queue.

{
  "frame": 1,
  "object": "object_01",
  "zone": "outside",
  "event": null
}

Select each timestamp to move the tracking box and inspect the corresponding event.

By the numbers

1,000

Deploys/day

50+

Agents

<1ms

Cold start

Challenge

Container sprawl at scale

Lumeo ran 50+ autonomous agents across 15 regions. Each agent needed its own container, sidecar, and orchestration config. Kubernetes overhead consumed 40% of their infrastructure budget and deployments took minutes, not seconds.

Solution

WASM-native agents on Hives

Lumeo migrated each agent to a WASM workload on Hives. The agents kept their existing logic but dropped containers entirely. Deploys went from 3-minute rolling updates to sub-second activations. The Nectar runtime provided sandboxing, capability gating, and memory limits without sidecars.

Result

1,000 deploys/day, 60% cost reduction

With Hives, Lumeo ships 1,000+ deployments per day with zero downtime. Infrastructure costs dropped 60%. Cold starts are under 1ms. The team spends time on agent logic instead of Kubernetes YAML.

From the team

“We deleted our entire Kubernetes config and replaced it with hive.toml files. Deploys went from minutes to milliseconds. Our agents are faster, cheaper, and simpler to operate.”

Elena Park

VP Engineering, Lumeo

Start building

Ship your first workload today. Same platform Lumeo trusts in production.

Get started →