AI accountability · Calgary, Alberta
Your team used AI today. Who approved the work?
Proposals, reports and client emails are being written with AI every day. When one goes wrong, most organizations cannot show who checked it or who said yes. AIWAA gives every AI-assisted task a named owner, a named reviewer and a permanent record.
Prepared with AI assistance
60-second exposure check
How exposed is your team right now?
Answer six questions honestly. Nothing is sent anywhere. Your result stays on this page.
What the research says
AI is moving faster than ownership.
of employees rely on AI output without checking its accuracy.
KPMG & University of Melbourne, 48,000+ people in 47 countries incl. Canada and the U.S., 2025
hide their AI use and present AI-generated work as their own.
KPMG & University of Melbourne, 2025
of companies reported financial losses from AI risks. The average was US$4.4 million.
EY Responsible AI Pulse, 975 C-suite leaders, Oct 2025
of senior data leaders at large companies cannot fully trace AI decisions.
Dataiku/Harris Poll, 812 data leaders, Aug 2025
of enterprises will demote or decommission AI agents by 2027 because governance gaps surface only after incidents.
Gartner, May 2026
Canada's rank among 47 countries for trust in AI systems.
KPMG & University of Melbourne, 2025
Canada · 2024
A tribunal held Air Canada liable for what its website chatbot told a customer.
The company argued the chatbot was responsible for its own words. The tribunal disagreed.
Moffatt v. Air Canada, BC Civil Resolution Tribunal
Australia · 2025
A A$440,000 government report with fabricated citations ended in a partial refund.
The revised report disclosed that AI had been used in writing it.
Associated Press, Oct 2025
AIWAA, explained simply
One office day. Two endings. Here's the difference AIWAA makes.
No technical words. Follow one client proposal from the moment the manager asks for it in the morning to the moment everyone goes home, and then three weeks later, when the client calls.
AIWAA in four steps
Every piece of AI work gets a name, a check and a record.
Assign
Who is doing it?The manager names who writes the work and who checks it.
Do COMING
AI helps write itThe draft is written with AI help.
Check
Did a person check it?The named person reads the exact draft and says yes or no, with a reason.
Record
Can we prove it?Who wrote, checked and approved is saved, with the exact final version.
Steps 1, 3 and 4 work in the prototype today. Step 2, where AIWAA's own AI writes the draft, is being built. Until then, your team uses the AI tools they already have.
The process map
From walking in to walking out.
Sara emails Priya: "Can you send the client proposal today?"
Nobody is named to check it. Everyone assumes someone will.
Sara opens AIWAA and creates the task.
"Client proposal. Priya writes it. Daniel checks it." Priya and Daniel each click Accept, so both know it's their job.
About 2 minutesPriya asks ChatGPT to write it. Done in two minutes.
It looks great. Hidden inside: a wrong savings number and a "2024 industry report" that doesn't exist.
Mistake hidingPriya writes the draft with AI help and sends it to Daniel inside AIWAA.
Same AI, same speed. The same hidden mistakes are in this draft too.
Mistake hidingPriya tidies the wording.
Nobody reads it for facts. There is no step that asks anyone to.
No checkDaniel reads the exact draft.
He spots the wrong number and the report that doesn't exist. He clicks Reject and writes why.
Mistake caughtBusy afternoon. The proposal sits ready to go.
The mistakes are still in it.
Priya fixes it. Daniel approves this exact version.
If anyone changes a word after that, it needs approving again.
Approved4:52 PM. Priya sends it to the client.
Wrong number and fake source included.
Mistake reaches clientSara clicks Release. Priya sends the approved proposal.
Correct, checked, and signed off by name.
Safe to sendEveryone goes home.
Nothing records who wrote it, who checked it, or which version was sent.
Everyone goes home.
AIWAA has kept the full story: who wrote it, who checked it, what was rejected and why, who released it, and the exact final version.
Record saved"Your number was wrong, and that report doesn't exist."
The CEO asks, "Who approved this?" Three people point at each other. Hours go on digging through emails. The client starts to doubt you.
Trust lost · time lost · blameThe client asks how the figures were checked.
The CEO opens the record in a few clicks: Priya wrote it, Daniel checked it and fixed two errors, Sara released it. The conversation is about facts, not blame.
Trust kept · answer readyThe office day is an illustration. Research: KPMG & University of Melbourne (2025); Moffatt v. Air Canada, BC Civil Resolution Tribunal (2024).
What the CEO gets
Six things you can count on.
A name on every piece of AI work
No more "I thought someone else checked it."
A person checks before it goes out
Mistakes are caught inside the office, not by your client.
Proof in a few clicks
When a client, auditor or board asks, you open the record instead of searching emails.
Staff use AI openly
People stop hiding their AI use when there's a fair, simple process to follow.
Fix the process, not the person
When something slips, the record shows where the process broke, so the team fixes it together.
A results report
The pilot measures mistakes caught, time to approval and time to find a record in your own team.
The risks it helps cover
What can go wrong, and how AIWAA helps.
| Risk | What it looks like | How AIWAA helps |
|---|---|---|
| Wrong facts or numbers | AI makes up a figure, a source or a quote, and it goes to a client. | A named person must check and approve the exact version first. |
| "Who approved this?" | Something goes wrong and nobody can say who signed off. | Every step is saved with a name and a time. |
| Your company is held responsible | In Canada, a tribunal held Air Canada liable for what its chatbot told a customer. | You can show that a human reviewed AI work before it went out. |
| Hidden AI use | 57% of employees admit hiding their AI use at work (KPMG, 2025). | A simple, fair process makes using AI openly the normal way. |
| Lost clients and reputation | A public mistake costs trust that took years to build. | Fewer mistakes reach the client, and you can explain any that do. |
| Audit and board questions | Leaders are asked to show how AI work is controlled. | A record you can export and hand over. |
How much could it save?
Put in your own numbers.
The numbers already filled in are only an example for a mid-sized team. Change them to fit your organization. The pilot then measures your real ones.
Your estimate per year
$0
This is a simple estimate from your own numbers, not a promise. It doesn't count bigger risks, such as a lost client or a legal claim, and it doesn't subtract the cost of AIWAA or the few minutes each check takes.
Plain and honest
What AIWAA is, and isn't.
It does not
- Make AI perfect. It makes sure a person checks it.
- Spy on staff or punish them. It's built to support people.
- Replace your lawyer or make you "compliant" on its own.
- Need IT or coding to start.
What happens next
Try it with one workflow.
- Pick one job where your team already uses AI, such as client proposals.
- We set it up and train your people.
- Your team works with it as usual.
- You get a report showing what changed.
Become a founding pilot partner · formativelabz.com
