Skip to content
DashboardStart deploying

System architecture

The full picture.

Five purpose-built layers. No container runtime, no orchestrator, no service mesh overhead. Just your code, sandboxed and deployed.

THE PLATFORM, CONNECTEDSee how the pieces fit.
01Your workloadWASM
02Runtime boundaryEXECUTION
03Connected servicesDATA + EVENTS

Architecture

See the boundary between every layer.

A request crosses four explicit layers, from your component to the services it can reach.

Interactive architecture example
Client requestResponse
Typed interfacesExplicit connections. Inspectable responsibilities.

Nectar runtime

Execution receives explicit capability handles inside an isolated memory boundary.

component → sandbox

Separate concerns

Control planePlacement & configuration
Data planeRequests & execution

Conceptual architecture, not a live topology.

Layer by layer

Five layers.
Zero bloat.

Every layer is purpose-built for WASM. No container runtime overhead. No Kubernetes complexity. Just your code, sandboxed and deployed.

Your codeLayer 1

Rust, Go, Python, JS -- compiled to WASM. Any language that targets the WASI 0.2 Component Model.

Nectar runtimeLayer 2

Wasmtime 40 execution engine. Snapshot restore, fuel metering, capability-based security. Sub-millisecond cold starts.

Queen schedulerLayer 3

Placement engine scoring capacity, proximity, load, cost, and compliance across 15 regions. Automatic failover.

Stinger functionsLayer 4

Serverless compute layer. HTTP handlers, event triggers, cron schedules. Scale to zero. Pay per invocation.

Data layerLayer 5

Cells object storage, Silos Postgres, Buzz queues, Pollen secrets -- all integrated via WASI capability handles.

Data flow

How data moves through Hives.

Client→StingerHTTPS request hits the nearest edge node and routes to your Stinger function.
Stinger→NectarFunction code executes in a WASM sandbox with fuel metering and memory limits.
Nectar→Cells / SilosData reads and writes go through WASI capability handles. No HTTP overhead.
Nectar→BuzzAsync work dispatches to queues. Consumers process in parallel with exactly-once delivery.
Waggle→StingerScheduled jobs fire at the specified cron interval. Leader-elected, single execution.

See it in action.

Deploy your first workload and watch the architecture work for you.

Start free →Platform overview

Hives is part of L1fe AI — infrastructure for autonomous intelligence.

Platform

  • Overview
  • How it works
  • Architecture
  • Runtime
  • Nectar
  • Queen
  • Comb
  • Stingers
  • Buzz
  • Waggle
  • Cells
  • Pollen

Solutions

  • AI agents
  • Edge functions
  • Scheduled jobs
  • Message processing
  • Webhooks
  • Real-time APIs
  • Multi-tenant SaaS
  • Fintech
  • Healthcare
  • Retail
  • Government
  • Startups

Developers

  • Docs
  • Quickstart
  • API reference
  • CLI reference
  • SDKs
  • Open source
  • Roadmap
  • Changelog
  • GitHub
  • Discord

Customers

  • Case studies
  • By industry
  • By workload
  • Partners
  • Technology partners
  • Solution partners
  • Become a partner
  • Community
  • Events
  • Champions

Company

  • About Hives
  • L1fe AI Inc.
  • Leadership
  • Mission
  • Careers
  • Press
  • Blog
  • Contact
  • Contact sales

Trust

  • Trust hub
  • Security
  • Compliance
  • Privacy
  • Responsible disclosure
  • SLA
  • DPA
  • Privacy policy

Big ideas. Good company.

Start building

© 2026 L1fe AI Inc.

PrivacyService agreementContact
Built for what comes next.