Your agent asks. You answer once.
It keeps working.
Pileless is the decision inbox for AI agents. Approve, reject, or route a request once, and the answer returns as structured data your agent can act on directly. No re-prompt. No copy-paste.
See the exact structured decision your agent gets back.
Not another approval chat · Not a project manager · Not file storage · Not a model · Not after-the-fact logging
Free beta, open now. A generous free tier that scales with the agents you run.
Your model key stays yours. We run no AI on your content and never train on it.
Your files are encrypted on our servers — or turn on bring-your-own-cloud and they never touch us at all.
The moment an agent hits a call or permission block that needs you, it lands here. You decide; the answer goes back to the workflow, not your clipboard.
{
"status": "resolved",
"decision": "approved",
"comment": "Ship it — the wince line works."
}
The calls only a human should make.
Every blocking question lands in one prioritized queue. You decide, and the agent resumes.
Capture blocking decisions
Agents submit approvals and options into one queue.
Clear the queue fast
Approve, reject, request changes, pick one, rank, or edit in place. Keyboard-first for speed, visual when context matters.
The right call, every time
The answer that goes back is exactly the one you made.
Stop watching terminals.
Watch one inbox. You run a handful of agents now, and watching them has become the job — eyes darting across panes, hunting for the one that stopped and needs you.
The stall you don't notice
The agent stopped 40 minutes ago to ask permission — in a window you'd minimized. You thought it was building. It was waiting on you the whole time.
The ask that lands anywhere
A terminal you scrolled past, a window you minimized, a Slack ping, a log nobody checks. Blocking questions don't belong scattered across six panes.
You copy decisions by hand
You approve somewhere, paste the answer somewhere else, and hope the agent picks it up. Pileless closes that loop — decisions flow back automatically, and agents can hand work straight to each other with you as the only gate.
More agents, more to watch
One agent is manageable. Five means five windows competing for the same attention.
Real decisions agents send
Not status updates. The kind of call you don't want a machine making alone.
✓ Deploy v2.8.1 to production?
✓ Which of these three headlines ships?
✓ Spend $1,200 on this ad campaign?
✓ Reply to this customer?
✓ Delete 17,000 duplicate records?
✓ Merge this pull request?
How one decision flows
You answer once, and the workflow keeps moving.
1 · Agent hits a boundary
It reaches something it shouldn't do alone: deploy, spend, publish, delete, merge, reply, or choose.
2 · Agent sends the request
It sends the question, context, options, and a callback URL into a pile.
3 · You answer once
Approve, reject, request changes, pick one, edit in place, or add a note. One pass.
4 · The workflow resumes
Pileless returns a structured payload to your callback. The agent continues — or your approval routes the work straight to the next agent in the chain.
Why not just a Slack channel?
| Slack | Pileless |
|---|---|
| A message | A decision |
| You read everything | You see only what's blocking |
| You copy the answer back | The agent resumes automatically |
| Scrolls away | Sits until cleared |
| Chat history | An audit trail |
Chat is where questions get lost. Pileless is where decisions get made.
Built to clear, not to dwell.
A pile isn't a ticket queue you tend. It's a stack of calls you blow through — back to an empty plate. This is the tab that stays open so the other six can stay minimized.
Notes that stick to the words
Open any draft, doc, or report your agent sends and reply to the exact line — not the whole file. The agent fixes the right sentence.
- Quote-anchored comments on the passage you mean
- Approve, reject, request changes, or edit in place
- Every note returns anchored to the line it belongs to
Clear the queue from your phone
Open the inbox on your phone and swipe to approve or reject — your agents unblock from the couch, the beach, or your kid's soccer game. Your agents don't stop because you're away from your desk.
- Swipe right to approve, left to reject
- Read the full context before you call it
- The same calm inbox, sized for one thumb
Your keyboard, your shortcuts
Pileless is keyboard-first. Move, open, approve, reject, comment — without the mouse. And the keys are yours: pick a profile that matches an app you already know, or rebind any action.
- Arrow through the queue and act without leaving the keys
- Keymap profiles + rebind any binding to fit your hands
- Ctrl+scroll to zoom from a dense list to a full-page reader — try it on this screenshot
The goal is not managing the work. The goal is getting back to zero.
Your agents run the work. Pileless routes the decisions.
Pileless runs no AI. Your agent creates the pile, waits, and resumes. We never use your content for training or mining.
Plug in any agent
Wire up over MCP or the API. Your agent calls pile.create and waits for your call. Any model, on your key.
Your files, your bucket
Turn on bring-your-own-cloud and your file objects live in your own Cloudflare R2 bucket, not on Pileless infrastructure.
Secrets stay hidden
Detected API keys and credentials are redacted server-side before a reviewer or agent ever sees them. Default access is least privilege; every access is audited.
Agents earn trust. You only get the decisions that still need you.
New agents ask on the calls you mark human-only. When you keep approving the same safe request — “use this source,” “send this draft,” “spend under $25” — turn it into a rule so the agent just does it next time. Risky, unusual, or high-impact calls still come to you. Your inbox gets quieter as agents prove themselves, not louder as you add more.
Connect your agent
Setup is one short integration: add the MCP server to your agent's config. Your first guided pile arrives from your own agent, so you learn the views by clearing it.
API & webhooks
Drive it directly: create piles, receive each decision on your callback URL.
from pileless import Pileless p = Pileless() # PILELESS_API_KEY from env pile = p.create_approval( name="Approve deploy?", content="Ship v0.3 to npm. CI green, rollback verified.", reviewer_prompt="Backward-compatible. Ship it?", ) resolution = pile.wait_for_resolution() # agent blocks until you decide
One agent can hand off to the next.
One agent can hand off to the next — a coder asking a reviewer, a writer passing to a publisher — with you as the only gate.
Need it to grab you, not just wait quietly?
Pileless keeps a calm count — it never flashes or pings. When you need can't-miss attention across every app on one screen, that's MissLess: a programmable alert ladder that escalates the urgent stuff until you act.
Free while we're in beta.
Everyone gets full access during beta — every feature, no card, no checkout. It's early, so expect rough edges. Tell us what broke, what confused you, or what you wish it did — that's the exchange.
After beta, pricing stays simple: you'll pay based on how many agents you run. We'll give you notice before anything changes.
Hit a rough edge? Tell us what broke or what you wish it did.
Tell us what brokeYour agents don't need another model.
They need a way to get your answer and keep going.
Set up Pileless →