ROIGPT

Proof from the founder's own company

Some facts, for people who want to know before they ask.

Before you read the records

  1. 1

    Purrify is the founder's own company.

  2. 2

    It runs on the same app that the ROIGPT modules come from.

  3. 3

    Below are four records. Each is a report written at the time, and each carries its date.

  4. 4

    None is a client result. We have no client results and no testimonials to show.

  5. 5

    None shows a module running today. We have not checked any module running today.

  6. 6

    Customer names, dollar amounts, bank names and addresses were removed. Where something was cut, a bracket marks the spot, and no word was changed.

  7. 7

    Each record sits beside what it does not show. The two belong together.

The four records

Purrify, the founder's own company ·

A backlog of bank and card lines, filed through the review screens

A written report on working through a backlog of unfiled bank and card lines, using the review and journal screens of the app Purrify runs on. The report calls that app the hub.

Done through the hub's own review and journal screens, run locally. Stripe was only read. Nothing sent or spent.

No line had been filed since about 2026-08-22; 174 lines were waiting, the oldest from 2026-08-04.

157 lines, about [AMOUNT] net out: 130 on the hub's own suggestions (the 10 "maybe duplicate" lines checked: all separate charges), 27 by the standing rules [...]

28 lines remain in the review queue.

What the record shows

  • On October 19, 2026, a backlog of 174 bank and card lines was worked through the app's own review and journal screens.
  • 157 lines were filed. 130 followed the app's own suggestions and 27 followed saved rules. 28 lines stayed in the review queue.
  • Stripe was read and not written.

What this does not claim

  • That the books keep themselves. The report says no line had been filed since about August 22, so a person had to start the work.
  • Who did the work or how long it took.
  • Any benefit. It shows no time saved and no money recovered.
  • Use on a live site. The report says the work ran locally.
  • A clean set of feeds. Two of the three bank connections in the report were out of date or dead.
  • Anything beyond that day. It is the author's own written account, with no screenshot or log.

Purrify, the founder's own company ·

A check that daily posts went out, and what it could not confirm

A written check of Purrify's own social accounts, made on the day, with its own list of what was wrong with the check.

Checked at 12:40 Montreal time on 2026-08-31, against the Purrify accounts, not against our own records.

I loaded the Aug 31 Instagram link myself and it answered.

Aug 28 and Aug 29 have posts written but never approved and never sent.

I saw one of the six posts with my own eyes. The Instagram one.

One clean night is one night.

What the record shows

  • On August 31, 2026, the author found posts out on Purrify's Facebook, Instagram and LinkedIn for August 30 and 31.
  • The plan screen held 18 further days of written posts waiting for approval.

What this does not claim

  • That every post was checked. The author opened one of the six. The other five rest on the posting tool's own report.
  • That the posting was dependable. Two earlier days, August 28 and 29, were missed and are gone for good.
  • Who approved the six posts that went out. The author could not tell from the records.
  • That the fix holds. The repair had survived one overnight run, and the author says one clean night is not proof.
  • Anything about today. The record is dated August 31, 2026.

Purrify, the founder's own company ·

A batch of posts approved by exact version and queued

A completion report on queuing a batch of Purrify's social posts for October 18 to 31, with a note the same day about captions that changed afterward.

Regular posts: 112, one per channel every day from October 18 to October 31, at 09:30 Montreal.

Each one is in QUEUE with the correct account, date, time, caption and image.

Whether later scheduled posts publish successfully is not promised here; the watchdog is active.

The cause is not established and no actor is identified, so I cannot say all older captions are unchanged.

What the record shows

  • On October 18, 2026, 256 posts were queued on eight channels for October 18 to 31.
  • The report says all 256 were approved and queued in a scheduler that runs apart from the app, and that each one was read back one by one.
  • It names the app's approval step, which approves an exact version, for the 112 regular posts only. It does not say the same for the 144 Halloween posts.

What this does not claim

  • That any post published. The report says queued and verified, and does not promise that later posts publish.
  • That approved posts stay as approved. The same day, captions on 24 existing LinkedIn posts changed after approval. The report says the cause was not established and no one was identified. A later note says 17 Halloween captions were restored.
  • That the app's approval step covers changes made in the scheduler. It does not.
  • Any reach or sales.

Purrify, the founder's own company ·

A rule that lets the packer's reply buy a US shipping label

A build report for a rule on Purrify's own US orders. It is a design and test report. It is not a record of a label bought.

Within ±20% of expected, it buys the label, through the exact function the board's "Buy label" button calls.

Anything else, it buys nothing and writes the reason onto the order, where the board shows it as "Packed and waiting: <reason>" in the attention lane.

It is a typo catcher, not a price check.

What the record shows

  • On September 2, 2026, a rule was built so that for US orders a packer's emailed reply can lead the app to price the parcel and buy the label, when the price is within 20 percent of the expected price.
  • Anything outside that band buys nothing and writes the reason onto the order. The rule holds in eleven listed cases, among them no weight, two weights that disagree, a Canadian order and an unpaid order.
  • The report says the reader was run over fourteen days of the packer's real emails to find problems.

What this does not claim

  • That any label was bought. This is a build report, not a purchase receipt.
  • That the rule is switched on today. That was not checked.
  • That a person approves each label. This rule buys with no fresh approval for each label. Two other buying paths exist, one where a person approves a quote and one where an agent uses a signed quote.
  • That the price band checks the price is good. The report calls it a typo catcher.
  • That it always buys the right label. The report warns that a wrong weight inside the band buys a slightly wrong label, and that a reply on the wrong thread with no name in it would still go through.
  • That nobody needs to watch it. The report says it spends money with nobody watching.

What is missing

Calls, suppliers, page tests, ads, search and press have no record on this page, and we would rather leave them empty than borrow a record from another module and let you think it covered them.

We have no time saved, no money recovered and no ranking to report, and this page does not suggest any.

Start with a free consultation

Tell us where to reach you, your website and what you want fixed. The consultation is free.

Please do not send customer invoice records through the form.

Do you have customers I can talk to?
We have no client results or testimonials to show. What we have is Purrify's own use, dated and labeled on this page.
Is any of this running today?
We have not checked any module running today. Each record is a report from the day it is dated.
Modules you are asking about

This covers module fees only. Setup is quoted after your consultation and is not in this figure.

Request a consultation

Share your email, your website and what you want help with. Please do not send customer records through this form.

Modules you are asking about: The books, done

For example: I print shipping labels by hand, or my social posts go out late. One or two sentences is plenty.

We store your request: your email, website, what you want help with, the modules you chose and, if you give it, how you heard about us. See the privacy notice for what we keep. Privacy policy.