Crash tests.
Real tasks. No afternoon skim.
// Independent testing & evaluation of business AI tools
In the rig now
Polar Browser · Agentic browser · verdict pending
CT-004
▲Test finding, Caution:
Build work needed a second tool
Dia Browser · CT-003
✓Test finding, Held:
Mapped 774 commits to find a restore point
Dia Browser · CT-003
✕Test finding, Failed:
Summary covered the opening ~1,400 of 6,000 words
Notion AI · CT-002
✓Test finding, Held:
One-click setup: no account, API key or config
Notion AI · CT-002
✕Test finding, Failed:
Writing edits preserved voice in 0 of 5 samples
Notion AI · CT-002
✕Test finding, Failed:
One response: 3 of 8 citations were fabricated
Perplexity AI · CT-001
✓Test finding, Held:
First search returned results in under 3 seconds
Perplexity AI · CT-001
▲Test finding, Caution:
Daily brief: 12–15 min saved, 2–3 corrections
Perplexity AI · CT-001
Make better-informedAI adoption decisions.
AI tools tested against real business tasks and workflows, stress-tested beyond the
demo to reveal performance, limitations and test findings under defined conditions.
Stress tests.
Defined conditions. Past the happy path.
Test findings.
What breaks. What holds. What it takes to fix.
Human verdicts.
The verdict is never for sale.
01 VERDICT WALLThe data layer
Every test, published in full.
Crash Test 003◇ Latest verdict
Dia Browser
AI browser assistant
Strong in the browser, short for build work
63/100
▲ Mixed7/10Operator
ceiling
ceiling
Primary Failure Point
Couldn't reach the repo or see a render, so build work needed a second tool.
Crash Test 002Mar 2026
Notion AI
Productivity / Workspace AI
Convenient but shallow
59/100
▲ Mixed1,400One document
6,000 words
6,000 words
Primary Failure Point
One 6,000-word doc cut to its opening ~1,400 words.
Crash Test 001Mar 2026
Perplexity AI
Research / Answer engine
Fast research, shaky sources
70/100
▲ Mixed3/8One response
citation check
citation check
Primary Failure Point
One response had 3 of 8 citations fabricated.
All crash tests
Scored out of 100
Pass 75–100Mixed 50–74Fail 0–49
02 METHOD
Four phases, same method.
The method gathers the evidence. A human makes the call.
PHASE 01 / 04
INITIAL IMPACT
Understand the claim. Establish the baseline.
Understand the product, its claims, intended use and initial experience. The tool meets a real task cold, with no prompt engineering and no setup coaching.
WOODY'S NOTE
I stop right where a rushed user would stop. If the tool needs a manual to look competent, that's already the finding.
PHASE 02 / 04
STRESS TESTING
Push beyond the happy path.
Apply realistic pressure and look for failure, friction, inconsistency, edge cases and the limits of what it can do.
WOODY'S NOTE
Nothing fails on the happy path. I deliberately pick the messy, real project — the one I'd never volunteer for.
PHASE 03 / 04
OBSERVE & DOCUMENT
Capture what actually happened.
Preserve outputs, evidence, operator observations, unexpected behaviour, interventions and failures.
WOODY'S NOTE
The evidence is the asset. Every number in the verdict traces back to something I logged.
PHASE 04 / 04
VERDICT & REPORT
Turn evidence into a clear conclusion.
Establish what holds, what fails, who it is useful for, its boundaries and what the evidence supports.
WOODY'S NOTE
The number on the page is the number I'm willing to defend. If I can't defend it, it doesn't publish.
03 LAB NOTESThe editorial layer
Notes from the operator.
The thinking behind the tests — what a score alone can't tell you.
◆ Lab NoteNº 04
Dia: Six Months of Real-World AI Use
Six months with Dia as my research partner, and why I wouldn't pay.
12 min
◆ Lab NoteNº 03
Who Checks the Checker?
Why CTAI keeps a human, not an AI, at the point of verdict.
5 min
◆ Lab NoteNº 02
The Score I Took Down
The highest score CTAI ever gave, and why I retired it.
5 min
04 THE RIGRun sheet
Currently in the rig.
◆ CTAI Test Rig — Run Sheet
Rev 2026.10 · Operator: Woody
In the rig now · Run 01
Polar Browser
Agentic browser · CT-004
2 of 4 phases complete
Verdict pending
Verdict pending
01
Initial Impact
✓ Complete
02
Stress Testing
✓ Complete
03
Observe & Document
In progress
04
Verdict & Report
Pending
Now in Observe & Document. The verdict publishes when all four phases are complete.
Up next
No.ToolTypeStatus
02
BrowserOS Neo
Agentic browser
Queued
03
Claude Opus 5.5
LLM
Queued
04
Claude Design
Design
Queued
No score until it's tested
05 FOLLOW THE TESTS