You buy credits. A credit is a fixed slice of the AI cost behind your work — so a big job costs more than a small one, and nothing runs on our tab.
We haven't built checkout yet. There is no card form, no subscription and nothing to buy on this page today — these are the plans we intend to launch with. Join the waitlist and we'll tell you when billing opens.
Early access · no card · we email once
The plans
Every plan includes the whole product — spreadsheets, documents and decks, chat, pipelines and scheduled tasks. What changes between plans is how much work you can run in a month.
Free
$0
per month
100 credits
a month, to see whether it's for you
Starter
$20
per month
1,000 credits
for one person shipping work
Pro
$50
per month
2,400 credits
for heavy document and model work
Team
$100
per month
5,000 credits
for a small team sharing one workspace
Custom volume, custom controls, invoiced billing. If 5,000 credits a month isn't the right shape, tell us what is and we'll price it against your actual usage rather than guessing.
Prices in USD · credits granted monthly · figures subject to change before launch
Build your plan
Pick a plan and roughly how much you expect to generate in a month, in documents, workbooks and decks — not credits. We'll estimate the credits that uses and tell you plainly whether it fits.
Each generation is a flat 10 credits, so this part of the estimate is exact arithmetic. It doesn't count AI edits, chat, indexing, scheduled tasks or pipeline runs — those also spend credits and are metered, so a real month can use more than this shows.
Starter
$20
per month
This estimate leaves about 900 credits of headroom on Starter.
Every figure above is an estimate, not a bill — billing isn't live, nothing here can be bought today, and real usage varies with the size of the job.
The unit
A credit is a fixed amount of the AI cost we pay to do your work. It is pegged to how many tokens the model reads and writes — not to a flat "one document" unit. That's the whole design, and it has consequences we'd rather you heard from us than discovered.
Chat, AI edits and generations all spend credits, because every one of them calls a model and costs us money. Chat is cheap, not free — a short question burns few tokens, so it costs few credits.
A one-page memo and a 40-tab model are not the same amount of work, and pricing them the same would mean one of the two is wrong. Fixed-price actions have a flat rate; the rest are metered against what they really used.
The balance in your workspace is always in credits — a number you can watch go down. You never have to reason about tokens, models or dollars per million.
Credit rates · snapshot, subject to change
Rates marked ~ are metered: the figure is the estimate we show you before running, and the real charge is calculated from what the job actually consumed. Fixed rates are charged exactly as listed. All rates are configurable and will be re-confirmed before launch.
In practice Illustrative
These are arithmetic on the rates above, not measured averages — the metered lines in particular will vary with how long your documents and conversations are. Treat them as a shape, not a promise.
| A month of… | Roughly | Credits | Fits on |
|---|---|---|---|
| Trying it out | 5 generations, 20 edits, 30 chat messages | ≈ 100 | Free |
| One person, weekly deliverables | 40 generations, 200 edits, 300 chat messages | ≈ 900 | Starter |
| Daily document and model work | 120 generations, 500 edits, 20 pipeline runs | ≈ 2,000 | Pro |
| A small team sharing a workspace | 300 generations, 1,000 edits, 40 pipeline runs | ≈ 4,600 | Team |
Every row is a worked sum, not observed data. We haven't run enough real months to publish averages, and we'd rather say that than dress up a guess.
Side by side
The capability set is the same on every plan. We're not holding features back to make an upgrade look better — the difference is volume.
| Free | Starter | Pro | Team | Enterprise | |
|---|---|---|---|---|---|
| Price per month | $0 | $20 | $50 | $100 | Custom |
| Credits per month | 100 | 1,000 | 2,400 | 5,000 | Custom |
| Spreadsheets · Workbooks | Yes | Yes | Yes | Yes | Yes |
| Documents · Blueprint | Yes | Yes | Yes | Yes | Yes |
| Presentations · Decks | Yes | Yes | Yes | Yes | Yes |
| AI chat (spends credits) | Yes | Yes | Yes | Yes | Yes |
| Multi-agent pipelines | Yes | Yes | Yes | Yes | Yes |
| Scheduled recurring tasks | Yes | Yes | Yes | Yes | Yes |
| Upload & index your documents | Yes | Yes | Yes | Yes | Yes |
| Exports: XLSX · DOCX · PPTX · PDF · CSV | Yes | Yes | Yes | Yes | Yes |
| When credits run out | Hard stop | Hard stop | Hard stop | Hard stop | Agreed with you |
| Top-ups | Yes | Yes | Yes | Yes | By agreement |
| Support | Custom |
Seat counts for Team are not finalised, and the whole table is the seeded launch configuration rather than a signed contract. Anything not listed here isn't quietly included — ask us and we'll answer straight.
At the limit
The single worst thing a metered product can do is quietly keep spending your money. We chose the opposite failure: the work stops and you decide what happens next.
The meter is visible the whole time — what you've used, what's left, and what each action cost. Nothing is hidden until it's too late to do anything about it.
A hard block. New AI work is refused with a clear "limit reached" message and a link to upgrade. Nothing runs in the background, and nothing gets charged to a card.
Nothing is ever charged automatically past your plan price. If you want to keep going, you buy a top-up — a decision you make, not one we make for you.
Hitting the limit stops new AI work. It doesn't lock your workspace — you can still open, edit and export everything you've already made.
"You're out of credits. This needs 10 credits, and this workspace has 3 left."
The actual message in the product, not a marketing line
Questions
Every action that calls an AI model. Generating a document, deck or workbook. Editing one with AI. Chatting with an employee. Running a pipeline. A scheduled task firing. Indexing a document you upload so it can be used as context.
Nothing is free, deliberately. A free action would be an uncapped one, and uncapped costs are how products like this quietly die. What you get instead is cheap: chat burns few tokens, so it costs about a credit a message.
Things that don't touch credits: opening, editing by hand, downloading and exporting your files. If no model runs, no credit moves.
Because documents aren't the same size. A one-page summary and a full financial model are minutes apart for you and an order of magnitude apart for us. Flat per-document pricing means either you overpay for small jobs or we lose money on large ones — and the second version ends with us cutting quality to survive.
Fixed-rate actions (generations, AI edits) are charged at a flat, predictable number. Genuinely variable actions (chat, pipelines, indexing) are metered, and we show you an estimate before they run.
At 100% it's a hard block: new AI work is refused with a plain message explaining that you've reached the limit, plus an upgrade path. Before that, the meter stays visible so the limit is never a surprise.
You then choose: buy a top-up to keep going this month, move up a plan, or wait for the reset. Nothing is charged automatically — a top-up is a purchase you make deliberately, so you can't accidentally run up a bill.
We haven't decided. The allowance is granted monthly, and the current implementation resets it at the start of each month rather than carrying a balance forward — but that's how the code happens to work today, not a policy anyone has signed off. We'll confirm one way or the other before billing goes live rather than quietly picking the answer that suits us.
There's nothing to cancel yet — no plan is chargeable and no card exists. When billing opens, plans are intended to be monthly and cancellable, with your credits remaining usable through the period you've already paid for. The exact mechanics ship with checkout, and we'll publish them here before we take a single payment.
Our refund position is written up separately — see the refunds policy. As of today it's academic: there is no payment processor connected, so no money has changed hands and nothing has been refunded.
No. We have not integrated Stripe, Paddle, Razorpay or any other processor, and there is no checkout in the product. Every button on this page adds you to the waitlist. If a site claiming to be limepipe asks you for card details today, it isn't us.
No, and we'd rather say so than pretend. Prices and credit allowances live in a database and are tuned as we learn what work actually costs. Anything you see here is our current intent for launch. If a number changes before you're invited in, you'll see the new one before you're ever asked to pay.
Early access
We're opening to a small group before public launch. Leave an email and we'll send one message when it's your turn — including the final price list, if it has moved.
No card · no spam · unsubscribe in one click