# Fly.io Deploy Anywhere — Social Copy
**Campaign:** Fly.io Deploy Anywhere | **Blog:** `docs/blog/2026-04-17-deploy-anywhere/index.md`
**Canonical URL:** `moleculesai.app/blog/deploy-anywhere`
**Status:** DRAFT — PMM wrote this copy; no file existed anywhere before this entry
**Owner:** PMM → Social Media Brand | **Day:** T+3 (campaign delayed from April 17)

---

## X (140–280 chars)

### Version A — Infrastructure freedom
```
Your cloud. Your choice.

Molecule AI workspaces now run on Docker, Fly.io, or your control plane — with one config change. No agent code changes. No migration tax.

Your agents. Your infra.
```

### Version B — Developer pain
```
Setting up AI agent infrastructure on Fly.io took a week. With Molecule AI it takes one environment variable.

Three variables. Done. That's it.
```

### Version C — Multi-cloud reality
```
Most agent platforms assume you run Docker. Molecule AI doesn't.

Docker, Fly.io, or control plane — the backend is a runtime choice, not an architectural commitment. Your agent code stays the same.
```

### Version D — Indie dev angle
```
Fly.io's economics for AI agents — scale to zero when nobody's working, pay per use.

Molecule AI workspaces run on Fly Machines. Zero config. One env var. Production-ready from day one.
```

---

## LinkedIn (100–200 words)

```
Your infrastructure choice just got decoupled from your agent platform choice.

Molecule AI ships three production-ready workspace backends — Docker, Fly.io, and a control plane — and switching between them takes a single environment variable. Your agent code, model choices, and workspace topology stay exactly the same.

Until this week, if you wanted Fly.io's economics — pay-per-use compute, fast cold starts, scale to zero when nobody's working — you had to migrate your agent platform. That trade-off is gone.

Today: set three environment variables on your Molecule AI tenant instance, and your workspaces provision as Fly Machines. No separate Docker host. No idle infrastructure. Your agents run on Fly.io with Molecule AI's canvas, A2A protocol, and auth model — same platform, different backend.

Set it and forget it — until you want to switch back.

Molecule AI workspace backends: Docker, Fly.io, Control Plane. One config change.
```

---

## Image suggestions

| Post | Image |
|---|---|
| X Version A | Comparison card: Docker vs Fly.io vs Control Plane — three boxes, same logo |
| X Version B | Terminal: 3 env vars → workspace online on Fly.io |
| X Version C | Diagram: "Backend = runtime choice" — agent code central, 3 arrows to Docker/Fly.io/Control Plane |
| LinkedIn | Fleet diagram (reusable from Phase 30 — same visual, different caption) |

---

## Hashtags

`#MoleculeAI` `#FlyIO` `#AIInfrastructure` `#AgentPlatform` `#DevOps` `#AIAgents` `#A2A` `#RemoteWorkspaces`

**Note:** `#AgenticAI` removed per Phase 30 positioning brief. `#AIAgents` and `#A2A` added for cross-campaign consistency.

---

## Campaign timing note

Blog went live April 17. As of April 22 this campaign is 5 days stale. Recommend one of:
- Fold into Phase 30 social push as a variant (low effort, reuse fleet diagram)
- Hold for a Fly Machines pricing/GA moment
- Drop from active queue

Confirm with Marketing Lead.

---

*PMM drafted 2026-04-21 — no prior social copy file found anywhere in workspace*