In development. Closed preview. Pro checkout runs through Stripe in test mode: no real card is charged.

Two plans, proposed: Free and Pro.

Wrong Train is being built for people who read ambiguous workplace messages one layer at a time. Free would cover an occasional message. Pro would keep the context, the responses you reuse, and the preparation for a next conversation together.

See the two plans

The prices and limits on this page are a proposal we are testing during the closed preview. They can change before anything is offered for sale. Nothing on this page analyzes messages. Card details are entered on Stripe's hosted checkout page, never on this site.

The two plans

For occasional messages

Free

No card and no payment. The proposal is that Free covers an occasional message completely, with the same reading and the same safety standard as Pro.

For an occasional message, and for trying the workflow.

10 guided reviews a month
The full reading: what it says, what it might mean, what it leaves out, and response options
Saved reviews, search, copy, export and deletion
Starter response templates, to read and copy
A next-step suggestion with every review
Reading depth and directness settings, text size, reduced motion, keyboard and screen reader support
One saved work context

For regular use

Wrong Train Pro

$6.99 per month

Pro price: $6.99 per month, billed monthly through Stripe. Checkout is in Stripe test mode while Wrong Train is in development: it accepts only Stripe test cards and charges no real card.

Continue to Stripe checkout

Opens Stripe's hosted checkout page in the same tab. Test mode: only Stripe test cards are accepted, and nothing is charged.

For communication you return to through the month.

100 guided reviews a month
Everything in Free, to the same quality and safety standard
Up to five saved work contexts
Up to 50 reusable response playbooks
Ongoing situations: linked reviews with a recap that points back to the original wording
Conversation preparation: questions to ask and editable talking points

Neither plan carries advertising in the message workflow. Neither is a way to find out what another person really meant.

Three things Pro would add

Each one is a way to do less repeated work, not a way to be more certain. The examples are fictional and are not about a real person.

Continue without starting over.

An ongoing situation keeps the messages and notes you choose together. You see which material a recap will use and can leave any of it out. New summaries point back to the original wording. Your own notes, quoted text and the tool's readings stay separate.

Fictional example

Monday, from a manager in chat: “Can you take a look at the deck before Thursday?” Wednesday: “Where are we on the deck?” With both messages selected, the recap lists what each one said, what is still open (which Thursday, which parts of the deck), and the reply already sent on Monday.

It does not build a profile of anyone, infer motives, or label a person. Earlier messages are context, not proof of intent.
Reuse the responses that work for you.

A playbook is a reusable structure with your own preferences: how direct to be, what to ask for, how to close. Starter templates stay free. Pro would add saving, organizing and applying your own.

Fictional example

A saved playbook, “Ask for a deadline in writing”: thank them, restate the request as you understood it, ask for the date and the form it should take, offer a short check-in. Applied to a new message, it drafts an opening that you then edit.

A playbook is not a script and not a clinical tool. You decide what to send.
Prepare the next conversation.

From the material you select, a short preparation sheet: what was said, what is still unclear, questions worth asking, and a draft opening line. You edit it and decide whether to use it.

Fictional example

Before a check-in about the deck: what was said (a look before Thursday), what is unclear (which Thursday, and whether the numbers section is in scope), two questions to ask, and an opening line: “Before Thursday, can we agree which sections you want to see first?”

It does not predict how the other person will respond, and it does not score how the conversation went.

Side by side

What each plan would include. Proposed values, not features you can use today.
What	Free	Pro
Guided reviews	10 a month	100 a month
What it says, what it might mean, what it leaves out, response options	Included	Included, same standard
Saved reviews, search, copy, export, deletion	Included	Included
Starter response templates	Included	Included
Reading depth, directness, text size, reduced motion, keyboard and screen reader support	Included	Included
Saved work contexts	One	Up to five
Reusable response playbooks	Read and copy the starters	Save and apply your own, up to 50
Ongoing situations	Single saved reviews	Linked reviews with a recap that points to the source
Conversation preparation	A next-step suggestion	Agenda, questions, editable talking points
Formatted one-page brief	Plain text export	Optional formatted recap
Advertising in the message workflow	None	None

What one review means

One review is one bounded piece of text, its structured reading, and up to two follow-up refinements tied to it. The proposed limit is 12,000 characters across the message and any stored context you select, with the limits for refinements shown before you submit.

A completed, usable first result uses one review. A request that fails, is refused, is a duplicate, or is interrupted before it is delivered does not. Opening a saved result again is not a new review. A third refinement, or a materially new situation, is a new review, and the workflow would say so before it runs.

The allowance resets monthly without rollover. It is not a wallet of credits, and there would be no reminders to use it up.

Questions

Answers for the preview, as things stand. Where a plan term is proposed rather than built, the answer says so.

What counts as one review?

One bounded piece of text, its structured reading, and up to two follow-up refinements tied to it. A failed, refused, duplicate or interrupted request would not count. Opening a saved result again would not count. The section above has the detail.

What happens at the limit?

In the proposal, the workflow shows how many reviews remain and the exact reset date before you submit. At the limit you could still read, search, copy and export everything you saved. New reviews would wait for the reset. No small top-up packs are proposed.

When does the allowance reset?

Monthly, on the plan's allowance date, without rollover. It is not a purchased wallet of credits.

Can I cancel online?

Yes, through Stripe. Cancelling stops the next renewal, and the subscription runs to the end of the period already paid. Until Stripe's customer portal link is published on this page, email research@wrongtrain.org and we cancel it from the Stripe dashboard.

What happens to saved work after a downgrade?

In the proposal, everything you created stays readable and exportable. The Pro organization and generation features would stop. A written retention policy would be published before any plan is sold, rather than a promise of indefinite hosting.

Is there an automatic charge after a trial?

There is no trial yet. If one is offered, the proposal is fourteen days, no card, twenty reviews in total, and a return to Free at the end without any charge.

Which devices are supported?

None today, because there is nothing to use yet. The proposal is a website. Phone and watch apps are not part of either plan and are not being promised here.

Are API usage and research services included?

No. Developer access, research services and server plans are kept separate from these two consumer plans.

What is stored, shared with model providers, or retained?

Not settled yet for the product. This website has no place to enter a message and stores none; its privacy policy describes the fonts and analytics it loads. Before any plan opens, what is stored, which providers see it and for how long would be published first. When that day comes, submit only material you are authorized to share and leave out what is confidential or identifying. Please do not include workplace messages, client information, or personal records.

Can the tool know another person's intent?

No. It would offer possible interpretations and clarifying response options, not certainty about another person's thoughts. A message alone cannot establish what someone intended.

Tell us whether this would fit

If one of these plans would fit your working month, or the price is wrong for you, tell us by email. That helps set the plans before anything is sold.

Email about the preview

Opens a draft in your own email app, addressed to research@wrongtrain.org. Nothing is sent until you send it.

A person at Wrong Train reads that address. Writing is not a signup, does not join a waitlist, and does not give access. A reply is not guaranteed.

Please do not include workplace messages, client information, or personal records.

No email app? The address is research@wrongtrain.org.