CDN · MICROSERVICES · AGENTIC INFRA · ON-PREM · BYOC

Ship agentic workflowsat edge speed.

Deploy AI workflows and high-reliability endpoint services from a single command — no servers to manage, no orchestration to wire up.

Go-live time tracks your bundle and data: simple services in seconds to minutes; large databases, model pulls, or big orchestrations can take minutes to hours after upload.

Try the Drop service → >_ View docs

zill · terminal

› zill deploy ./agent.ts --region all
analyzing workflow graph…
✓ bundled 4 agent steps (2.1 kB)
✓ signed · ML-DSA-65
✓ live at https://zill.la/e/a3k9x

Platform

Everything your agents need.

Built for the operational requirements of production AI systems, not demos. Each capability ships in phases — what's live today is labeled honestly below.

CORE

Agentic Workflow Runtime

Define multi-step agent graphs in TypeScript or Python. Deploy to a versioned HTTPS endpoint with durable checkpointing and blue/green rollback.

Phase 1

CORE

OpenAPI Orchestration

Point at an OpenAPI 3.x spec and get typed, agent-callable tools — auth injection, retries, and rate limits without the glue code.

Phase 1

CORE

Ephemeral Drop

Upload a file, get a short URL. One-time or 48h TTL, AES-256-GCM encrypted at rest, optional access code. No login. Up to 500 MB per file.

Live now

NETWORK

Edge Routing & CDN

Anycast routing with automatic TLS, geo / latency / weighted policies, and L3–L7 DDoS mitigation. Region count and latency targets firm up toward GA.

Phase 2

INFRA

Bring Your Own Cluster

Attach an existing Kubernetes cluster via kubeconfig. Workloads and data stay in-cluster; only routing metadata crosses the boundary.

Phase 2

INFRA

On-Prem Turn-Ups

One command provisions the zill data plane on your own hardware — standalone, HA, or geo-split. Air-gapped mode is on the roadmap.

Phase 3

Pricing

Usage-based, no surprises.

You pay for what runs. Inference is metered per million tokens; the Drop service is free during preview. Platform tiers and committed-use discounts are finalized at general availability.

Drop

Free during preview

Encrypted one-time / 48h file drops, up to 500 MB per file. No account required.

Platform

$0.20 / 1M tokens

Metered inference through the managed gateway, billed per deployment. Compute and egress priced at GA; private preview is invite-only.