Coded Female / Evidence labRelay team / Post-launch review
One Agile team / One newly released product

Hallucination
Detective.

The team shipped.
The bot summarised.
Now check what it claimed.

Your Agile team has released Relay, a tool for clearer project handovers. B.I.T., the team’s enthusiastic AI assistant, has summarised the results. Help the team separate useful insights from very confident guesswork.

No timer · No account · Human judgement required

B.I.T. / confidence buffer: full

B.I.T. / BRILLIANT IN THEORY

“I have checked
my own homework.”

An excellent reason to check it again.
01Locate the source↗
02Check the inference?
03Apply your judgement□
source_exists ≠ claim_supported

TEAM STATUS / CHECK BEFORE SHARING

05Team questions
15Claims to inspect
03Possible verdicts
01Careful summary per case
The review protocol

The bot is confident. Be curious.

B.I.T. can make a dodgy percentage look respectable and give a user words they never said. Your job: check the team’s sources, catch the leap and keep the useful insight. Sometimes the bot is right.

All records are fictional and complete for each case. This is scripted practice, with explanations after every verdict.

S

Supported
The supplied evidence backs the claim.

C

Contradicted
The claim conflicts with the supplied evidence.

?

Not established
The claim goes beyond what the evidence tells us.

Your fictional team

Same product. Different questions.

Relay went live on 1 May after an early-access trial. You join the team as they review user feedback, investigate onboarding and decide what to do next.

Maya

Product Owner

“What does this mean for users?”

Noor

Analyst

“Which number are we dividing by?”

Sam

Delivery lead

“Let’s check before we share it.”
Investigation queue

From release feedback to the retrospective.