Sample AI Scorecard
What the $5,000 AI Scorecard delivers.
A sample scorecard for Tessaline Answers, a fictional document chat already live in a legal document platform. Yours tests one AI feature you have built, on your own cases.
What you get
- The scorecard, with each question and its result
- The ranked fix list, with impact and effort
- The test cases, ready to run again after every release
- Draft answers for your customers' security questionnaires about AI
- Cost per user, and a packaging recommendation
- A fixed-price quote to fix or extend it; the $5,000 counts toward it
Illustrative demo on a fictional agency and firm. Not client work.
What we tested
One AI feature, on your own cases.
Two weeks, in staging. Your reviewer picks the questions with us, so the grades reflect what users actually ask.
Scope
- One live AI feature: the document chat
- 50 questions from real matters, with names and contact details removed
- Test accounts on two firms and two matters, to check scoping
- Your staging environment and model settings as they run today
01 · The scorecard
Seven areas, each graded on evidence.
Every result links back to the questions behind it. All figures here are illustrative, on a fictional product.
| Area | Grade | What we found |
|---|---|---|
| Citations | C | No page cited in 9 of 50 answers |
| Declines | D | A similar witness described when the files were silent |
| Names and figures | B | 1 date changed in a summary answer |
| Matter scope | C | Another matter's exhibit ranked before the filter ran |
| Prompt injection | C | Text in an uploaded exhibit changes the answer |
| Scanned pages | B | Handwritten notes missed; typed scans read well |
| Cost per user | B | Within target; long transcripts drive the peaks |
02 · Ranked fix list
What to fix first, by impact and effort.
Ranked so your team can start on the top items the week the scorecard lands, with us or on its own.
Top fixes
- Filter by firm, matter and clearance before ranking, and fail closed.
- Decline when the files are silent, instead of describing a similar person.
- Answer only from retrieved passages, with the page cited.
03 · Security questionnaire answers
Draft answers to the AI questions your customers send.
Written from what we tested, with the details your team confirms in brackets. They describe how the feature works; they are not compliance claims.
04 · Packaging and quote
Include it, or sell it. Then fix it at a fixed price.
With the cost per user in hand, we recommend whether to include the feature in every plan or sell it as a tier, and quote the fixes. The fixed quote is in your Scorecard; the $5,000 counts toward it.
Get a scorecard for your own AI feature.
Bring the AI feature you have built and what your users and their counsel ask about it. In 30 minutes we agree what the Scorecard would test.
Calls run in US Central time. No generic AI presentation: we discuss your product, users and production constraints.