Invoice totals now round differently
- Plan
- Round each line item
- Code
- Rounds the final total
Shipped with Discount codes. Also affects: Tax report, Refunds
Ship fast.
Know what you shipped.
Truing checks what your AI agents built against what you planned, and asks you to decide when they don’t match.
Works with your Claude or ChatGPT subscriptionFrom GitHub or your own computer
Board · Acme Billing
Inbox · 3 need your call
Example product and data, for illustration.
AI made building cheap. Knowing what got built is the hard part now.
-
Day 1
You plan it
Add discount codes. Invoices keep rounding each line item.
-
Day 2
The agent builds something slightly different
Discount codes work. Invoice totals now round once, at the end.
-
Day 3
Nobody notices
You approve the change. It looks fine, and it’s one of dozens this week.
-
Day 17
A customer does
Their invoice doesn’t match their tax report. Now it’s a support ticket.
Multiply that by every change your agents ship. The plan goes stale, and nobody can say for sure what the product does.
Stay on top of everything your agents ship.
See your whole product
Every part of your product is a card on one board, in lanes by who uses it. Each line on a card shows where in the code it came from.
Know what a change affects before you make it
Describe a change in plain words. Truing shows which cards it touches, and why, before anyone builds it.
Decide where plan and code disagree
Truing checks the code against your plan and puts each mismatch in front of you, with both sides shown.
Ask about your product in plain words
Ask what the product does today, or what was decided and when. Answers come from your board.
Three steps to get you back in control.
-
Connect
Connect GitHub or your own computer, and add your PRD if you have one (Word, PDF or text). Use your Claude or ChatGPT subscription, or your own key.
-
Truing builds your board
It reads the code, writes a card for every part of your product, and compares what is there with what you planned.
-
Update plan Fix code Dismiss
You decide
Each mismatch becomes a question on its card. Every answer is saved, and the fixes you choose become a brief for your coding agent.
Set up in one sitting. The first cards appear in a few minutes, and you watch the board fill in.
Your coding agent knows the code. Truing knows the plan.
It remembers what you decided, and asks before anything changes it.
| Asking your coding agent | Truing | |
|---|---|---|
| What it remembers | Nothing between sessions | Your plan and every decision, with each version kept |
| When the code changes | You ask again, and it starts from nothing | Reads the changed code again, against the same plan |
| When code and plan disagree | You have to spot it yourself | You get a question on the card and make the call |
| Your decisions | Lost in chat history | Recorded, so you can see what you approved |
| What it covers | The files you point it at | The whole product, for everyone who uses it |
Truing works with your coding agents, not instead of them.
You choose where your code is read.
On your own computer
Connect your computer and Truing reads the code there, through your own Claude or ChatGPT subscription.
Or from GitHub
Connect GitHub and pick the one repo to share. Read-only access is enough.
Only you change the board
Code and documents raise questions. Nothing on your board changes until you decide.
FAQ
Questions teams ask first
How long does setup take?
One sitting. From a PRD, a couple of minutes. From code, the first cards appear in a few minutes and the whole board takes longer for a large codebase. You can watch it fill in.
Do I need a PRD?
No. Truing builds your board from the code alone. If you have a PRD, add it and Truing checks the code against it.
Where is my code read?
You choose. With your computer connected, the code is read there and the AI work goes through your own Claude or ChatGPT account. With GitHub and your own key, Truing reads the one repo you shared. Your board, with its cards and decisions, is kept by Truing so you can come back to it.
Which coding agents does it work with?
Any of them. Truing turns the fixes you choose into a brief you can paste into Claude Code, Cursor, Codex or another agent.
How much does it cost?
The AI runs on your own Claude or ChatGPT subscription, or your own key.
Who is it not for yet?
Large companies that need single sign-on, and products with very few user-facing parts. You don’t need to read code to use Truing, but you do need a screen larger than a phone.
Early access
See what your agents actually shipped.
We’re opening Truing to a few people every day.
- We set it up with you on your own code
- We walk through your first board together
- You tell us what Truing got right and wrong