Slotflow alternatives.
Slotflow does what it says — one API call to check availability and book, atomically, with a 409 on conflict. The reason developers outgrow it is coordination: Slotflow is strictly first-come-first-served, with no agent identity, no priority, no holds, and no audit trail. AgentDraft is the alternative built for the multi-agent case, where first-come-first-served is not coordination.
Updated
Slotflow (launched March 2026) got the hard part right: booking is atomic, and a lost race returns a 409 instead of a silent double-commit. That alone puts it ahead of agents writing straight to Google Calendar, where two agents can both pass an availability check and both succeed. The surface is genuinely minimal — "one API call to check availability and book" is an accurate description, not marketing — and the free tier's 100 bookings are enough to evaluate seriously. For a single agent doing simple check-and-book, there is nothing wrong with it.
Slotflow resolves every race by arrival order: whichever request reaches the server first wins. With one agent, that's fine — there's nobody to lose to. With several agents on the same calendar, it means the winner is chosen by network timing, not by anything you decided. Your background focus-blocker can beat the exec assistant's booking to a slot because its retry loop fired 40 ms earlier, and nothing in the system knows or cares which agent mattered more.
The missing primitives compound. No agent identity means the 409 can't tell the loser who won or whether renegotiating is worthwhile. No holds means an agent can't claim a slot while it confirms with a human — it must book immediately or risk the slot vanishing mid-conversation. No bump window means a mistake by a low-stakes agent can never be corrected by a higher-stakes one. And no audit trail means that when the calendar looks wrong on Friday, there is no record of which agent did what, in what order, and why. There's no mailbox either — the mail that triggered the booking lives in a different vendor entirely.
| Comparison point | Slotflow | AgentDraft |
|---|---|---|
| Per-agent email inbox | No | Yes — addressable per-agent mailbox, audit-logged |
| Conflict-free calendar writes | Yes — atomic booking, 409 on conflict | Yes — storage-level conditional write; one row per time bucket in one transaction |
| Agent priority + bump window | No — strictly first-come-first-served | Yes — per-user ranking; 30s default bump window |
| Holds with TTL | No | Yes — claim a slot while confirming, auto-expires |
| Append-only audit trail | No | Yes — across mail, holds, commits, evictions, rule changes |
| Typed 409 on lost races | 409 on conflict; no winner identity or priority | Yes — outranked with winner's identity and audit reference |
| Entry price | Free (100 bookings); paid $79/mo → $299/mo | Free Developer tier, no card; $10/mo Individual; $25/mo Team |
Slotflow rows reflect the public product surface as checked on 2026-07-01. If we've mis-stated a feature, tell us and we'll correct it.
Genuinely: if you run a single agent, the flow is check-availability-then-book with nothing in between, and you don't need email or multi-agent coordination, Slotflow is a reasonable fit. Its atomicity guarantee is real, the API is small enough to integrate in an afternoon, and priority ranking you don't need is complexity you shouldn't pay for. A one-agent booking bot behind a form does not need a coordination layer — and if that's the whole roadmap, switching buys you little.
Choose AgentDraft the moment a second agent shows up on the calendar — yours, a vendor's, the user's own assistant. The conflict engine resolves races by an explicit per-user priority ranking rather than arrival order: in the open-source collision benchmark, the rank-1 agent won 100.0% of races at p99 112 ms across 500 concurrent attempts, with 0 double-commits. Holds with a TTL cover the confirm-with-a-human gap, the 30-second bump window lets a higher-priority agent correct a lower one, every state change lands in an append-only audit log, and each agent gets its own email inbox — so the mail that triggered the booking and the booking itself live behind one API.
Frequently asked
Doesn't Slotflow's atomic 409 already prevent double-booking?
For a single contested write, yes — both products guarantee one winner per slot. The difference is which writer wins and what the loser can do next. Slotflow picks whichever request arrived first, regardless of which agent matters more; AgentDraft picks by an explicit per-user priority ranking, and the loser's 409 carries the winner's identity so it can renegotiate instead of guessing.
What does "first-come-first-served is not coordination" mean?
First-come-first-served resolves a race by network timing: the fastest request wins. Coordination resolves it by policy: the agent you ranked highest wins, a committed booking can still be bumped by a higher-priority agent inside a 30-second window, and every decision lands in an append-only audit log. With three or more agents on one calendar, timing-based outcomes are effectively random.
Can my agent hold a slot while it confirms with a human on Slotflow?
No — Slotflow has no hold primitive, so the agent must either book immediately or risk losing the slot while it waits for confirmation. AgentDraft has holds with a TTL: the agent claims the slot, asks the human, and either commits or lets the hold expire. The hold occupies the slot in the conflict engine, so no other agent can take it in the meantime.
How do Slotflow and AgentDraft prices compare?
Slotflow starts free with 100 bookings, then paid plans run $79/mo to $299/mo. AgentDraft's Developer tier is free with no card (one agent, one mailbox, 50 bookings a month), Individual is $10/mo for up to three agents, and Team is $25/mo with shared priority resolution. If you need more than one agent, the price gap widens in AgentDraft's favor.
- AgentDraft vs. Slotflow — the head-to-head comparison, feature by feature.
- The multi-agent collision benchmark — the 500-attempt receipts behind the coordination claims.
- AgentDraft pricing — free Developer tier, $10/mo Individual, $25/mo Team.
- Multi-agent scheduling glossary — bump window, holds, outranked, and the rest of the vocabulary.