How it works
From one sentence to a booked job
You say what should happen. PineMogul checks it against your own data, shows you one plan card and changes nothing until you say yes. Here is every step in between.
How it works
You say what should happen. PineMogul checks it against your own data, shows you one plan card and changes nothing until you say yes. Here is every step in between.
A scripted demo of the chat. Pick a language, then answer Yes, No or Change, just as you would in the app.
Pick a language, then answer the card: Yes, No or Change. A scripted demo: in the app you type anything, in any language. Example company, invented for the demo. Its amounts are in its own currency, Macedonian denars.
Write it the way you would tell a colleague: "Tomorrow 2 workers to Oak Street with the mini excavator, and charge Brightwater Homes $1,800." No forms, no menus.
When, how many people, which site, which gear, what price, which client. Everyday commands are read by our own code at no AI cost. The AI model is called only when that code is not sure.
The client and the site, who is free tomorrow, whether the gear is free, your usual costs. These are exact lookups in your own records, never guesses by a model.
Every step runs as a dry run inside the database and is then rolled back. The card only shows what would succeed right now. If a step would fail, you get the reason instead of a card.
One plan card with crew, gear, price, cost and margin, and three answers: Yes, No or Change. Up to this moment nothing has been touched.
Your yes runs the whole plan in order. A double tap or a shaky connection never creates two jobs. Undo stays open for ten minutes.
This is the card for the sentence above. Every figure on it comes from your own records and is written by our code, not by the AI.
Tomorrow 2 workers to Oak Street with the mini excavator, and charge Brightwater Homes $1,800.
Create this job?
Yes creates the job, books the crew and the excavator and sets the price. The chat answers "Done. Job #1042", linked to the job in the dashboard, with Undo right next to it.
Yes runs the plan. If someone booked Mark in the meantime, the step that is no longer possible is refused, nothing of the plan is left half done, and you get a fresh card built on the new situation. An old card never runs a plan you can no longer see.
No closes the card. Nothing was changed, so there is nothing to clean up.
Change is for corrections: another crew member, another day, a different price. You say what to change and get a new card to check. A new card always replaces your older open ones, so a forgotten yes cannot run an earlier plan.
Undo is there for ten minutes after a yes, for the person who said it. It reverses every step of the plan, in reverse order, all or nothing.
These rules are enforced by code, with the rights of the person who is asking, not by instructions to the AI. A foreman cannot approve payroll by asking nicely.
"Who is free tomorrow?" or "Which jobs are open for Brightwater Homes?" Answered straight away. Nothing changes.
Create or move a job, assign people, reserve gear, add a client. A plan card, then your yes, with undo afterwards.
Sending an email, for example. It runs on an admin's yes. Anyone else's yes sends it to the approval feed, for a person who is allowed to decide.
Deleting or voiding a record. Always on a card of its own, with a typed reason and the right role. Never automatic.
Not everything starts in the chat. Requests from your team, emails about to leave and notices about your account land in one feed, each as a card you answer with one tap.
Tap Yes or No, or press Y / N. Example company, invented for the demo. Its amounts are in its own currency, Macedonian denars.
When someone without the right to send or spend says yes in the chat, the request waits here for a person who has that right. Your approval rules decide who that is.
An approved email waits ten seconds before it leaves your mailbox. One tap stops it. More on mail and approvals.
"Your Microsoft login expired. Reconnect?" or "Storage 80% used. Add 50 GB for $5 a month?" Each notice appears once, not every hour.
New cards reach you in the app and as notifications, and each one opens on the record it is about.
Everyday job commands in English, German, French and Macedonian (Cyrillic or Latin letters) are understood by our own code, at no AI cost.
What our code cannot read goes to the AI model, and the answer comes back in the language you wrote in.
Hold to talk on the phone and on the web, with the same plan card coming back. Not available yet.
The app's screens are in English today. Macedonian, German and French are coming.
The chat has no back door into your data. It calls the same commands as the buttons in the dashboard, with the same rights and the same checks, so what you can say and what you can click never drift apart. A test in our build fails if an action in the app has no matching command.
Every answer links to its screen. "Done. Job #1042" opens that job, and the classic dashboard, with the dispatch board, invoices and reports, stays complete. You never have to chat.
No. The AI only proposes a plan. Our code checks it against your rules and the rights of the person asking, rehearses it, and runs it only after a yes.
The plan runs once. A repeated yes on the same card returns the first result instead of running the plan again.
Then the plan does not run as shown. Nothing of it is kept, and you get a new card based on the current data.
Only when the AI model is needed. Questions and everyday commands that our own code reads cost nothing. See AI included for the gauge and the fair limits.
Every plan starts with a 14-day free trial without a card and has a 30-day money-back guarantee.