Build · Schedule · Run

build

agents

that work

A lightweight workshop for AI agents. Write a prompt, set a schedule, press run. No subscriptions, no rituals — pay only for what your agents actually consume.

No credit card required · Cancel any time by not running anything

Daily News Digest
Running · 04 / 12 steps
01
Fetch sources
6 RSS feeds · 3 newsletters
✓
02
Score relevance
42 → 12 items
✓
03
Cluster topics
4 clusters identified
✓
04
Draft summary
Writing 480 words…
05
Send to inbox
—
Schedule Daily · 07:00
Last run Yesterday
Cost / run ≈ $0.04

How it works

Three steps from idea to scheduled agent. The whole loop fits on one screen.

01

Describe

Tell us what your agent should do in plain language. We turn it into a structured system prompt you can refine.

02

Schedule

Run on demand, every hour, every Monday at 9am, or only when triggered. One schedule per agent — clear and simple.

03

Run

Watch it work step by step. Inspect every call, every output, every cent spent. Fork or revise any time.

Pay only
for what runs.

The usual way

Subscription

$20 /month, every month
  • You pay whether you use it or not
  • Tier limits surprise you mid-month
  • Hidden margins on every model call
  • Cancel-flow, retention emails, the whole thing
Our way

Pay as you go

$5 buys roughly 200 agent runs*
  • Top up when you need it. That's the whole pricing page.
  • Every run shows you the exact cost — to the cent
  • Pick any model — fast & cheap, or slow & sharp
  • No tiers, no seats, no surprise upgrade pings

* Average cost of a daily-digest agent on a fast model. Heavier agents run longer; we show you before you commit.

Light agent (lookup)
≈ $0.01
Daily digest
≈ $0.04
Lead scrape (50 sites)
≈ $0.18
Mood-board researcher
≈ $0.32
Deep research (1h)
≈ $1.20

The whole
workshop.

Six pieces, designed to fit in your head. No plugins to install, no workspace to onboard, no certification to earn. Open the dashboard, describe what you want, ship it.

01 Prompts

Guided system prompts

A conversational prompt-builder asks the right questions and writes a structured system prompt for you. Edit the result directly any time — nothing is hidden.

02 Schedules

Schedules that read like a sentence

"Every weekday at 7am". "Every two hours, only between 9–6". "When my inbox gets a Notion mention." Plain language → cron, with a preview of the next 10 runs.

03 Models

Every model. One bill.

Hundreds of models behind one wallet — frontier, fast, cheap, open-weight. Switch per agent or per step. Costs displayed before you click run.

04 Runs

Every step, in plain sight

A run is a transcript: inputs, tool calls, outputs, exact tokens, exact cents. Replay it, fork it, share it. No black boxes, no "trust us".

05 Community

Templates from people like you

Share an agent as a template; clone what others share — full instructions visible before you commit. Curated, credited, one click to a working copy.

06 MCP

Your agents, inside Claude

Connect Claude Code or Claude Desktop via MCP and drive your agents conversationally — list, run, collect results. One token from Settings, billed like any run.

Agents in the wild.

Three quick blueprints to fork. Open one, swap a few parameters, and you've got your own running in under a minute.

Daily

News digest

A morning brief from your sources, clustered by topic. 480 words, in your inbox, every weekday at 7am.

  • Daily 07:00
  • 5 sources
  • ≈ $0.04 / run
Fork this agent →
On demand

Lead scraper

Drop in a list of company URLs — get back a clean spreadsheet of decision-makers, with confidence scores per row.

  • On demand
  • Up to 200 rows
  • ≈ $0.18 / run
Fork this agent →
Weekly

Mood-board researcher

Give it a brief, get back 60 candidate images with sources and a one-line rationale per pick.

  • Weekly · Mon
  • Image search + tagging
  • ≈ $0.32 / run
Fork this agent →

Build your first agent
in the next ten minutes.

$5 free credit on signup. No card. No subscription. No countdown timer.