Features
AI Assessment Builder
The Scorvia AI assessment builder drafts an assessment from a description of who it is for. You name your industry and specialization, the AI writes a first-draft question set in that field's vocabulary, and you edit, score, branch and reorder it like a document. Add your own questions, reuse ones from your library, and publish when you are ready. Nothing goes out until you say so.
- 14-day trial, no card
- The AI drafts only when asked
- Four question types, video included
- No model trained on your content
The question types
- 01
Multiple choice
- 02
True / false
- 03
Open text
- 04
Video
- Drafted by specialization
- Four question types
- Video answers
- Scored options
- Conditional branching
- Your question library
- Drag and drop
- Preview before publish
- No template library
- No emotion detection
- Nothing added without you
Most assessment tools give you a blank form and a library of somebody else's templates. The blank form is honest but slow. The templates are fast but never quite about your clients, so you spend the time saved rewriting them.
The builder starts from the other end. It asks who the assessment is for and writes the first draft accordingly.
The draft
How the AI drafts your questions
Set your industry and pick every specialization you work in. The AI writes each set of questions to the specialization it belongs to; in the vocabulary those clients use.
That last part is the point. A performance-management set asks about calibration, review cadence and how underperformance conversations go. A stress-management set asks about recovery, sleep debt and workload boundaries. A wealth-management set asks what the client did in the last downturn, not how they rate their risk appetite out of ten. None of it is a generic template with the nouns swapped, because the draft was written for that field from the start.
What comes back is a draft you own. Keep what fits, rewrite what nearly fits, delete the rest. The AI writes when you ask it to and at no other time - nothing is added to your assessment on its own.
Name the specialization, and the questions arrive written. Then it works the way you would expect a document to work.
The set
Four question types
| Type | What it is for | Scored? |
|---|---|---|
| Multiple choice | Single or multi-select. Give any option a mark. Multi-select sums the marks, which is how you weigh a question where two of five answers are both partly right. | Yes |
| True / false | Quick, binary, and the cleanest trigger for branching. | Yes |
| Open text | Where the client needs room. The answer feeds the report in their own words. | No - a paragraph is not a number |
| Video (new) | The client records an answer which is transcribed. For questions people answer better out loud than in writing. | No - the transcript feeds the report |
New
Video answers
Some questions are better answered out loud.
Ask someone in writing what a bad week looks like and you get two lines, because typing is editing. Ask them to say it and you get the context, the caveat and the things they had not planned to mention. A client records their answer to the question, the recording is transcribed, and the words feed the report draft the same way typed answers do.
Video is transcription only. Nothing analyses facial expression, tone of voice or emotional state. The AI reads the words, and you read the words. Video is selectable per question, and a client can type instead.
Marks
Scoring where it makes sense
Give any multiple-choice or true/false option a mark, and the report carries the scores and the charts. Multi-select questions sum their marks. Open text and video stay unscored, because turning a paragraph into a number would be inventing a measurement.
Scores are yours, not standard. There are no validated norms and no percentile bands - a total means what your question design means. For a coaching diagnostic that is the right shape; for a licensed psychometric it is not, and the builder is not trying to be one.
The path
Conditional branching
A multiple-choice or true/false answer sends the client down a different path. Someone who says they manage a team gets the management questions; someone who does not never sees them. One assessment can serve a first-time manager and a director, or a new client and a returning one, without either being asked something irrelevant.
A twelve-question assessment that adapts beats a forty-question one that does not - and completion rates are where the difference shows up hardest.
Reuse
Your own question library
The questions you have used before are offered back to you for the specialization you are building. Drag one straight in rather than writing it twice. Over a few months this has become the most valuable thing in your account: a set of questions that have already worked, in your words, ready to reuse.
It is your library, not a shared one. Nothing another coach has written appears in it, and nothing you write appears in theirs.
The build
How you build one
- 01
Name your specialization
Set your industry and the specializations you work in. HR, finance and wellness ship configured; define your own for anything else.
- 02
Let the AI draft the questions
The AI writes a first-draft question set in the vocabulary of that specialization. It writes only when asked and never adds questions on its own.
- 03
Edit the set
Keep, rewrite or delete any question. Drag in your own multiple-choice, true/false, open-text or video questions, or pull ones from your library.
- 04
Score and branch
Give multiple-choice and true/false options marks. Route clients to different questions based on their answers so nobody is asked something irrelevant.
- 05
Preview, then publish
Read the whole set back with its scoring and branching laid out. Fix what reads wrong, then publish and invite a client by name and email.
Your hands on it
Edit everything, then preview, then publish
- 01
Drag and drop.
Reorder, edit, delete. Drag a blank question of any type into the set and write it yourself.
- 02
Edit any AI question down to a word.
The draft is a starting point, not a locked template. Plenty of coaches end up with half the set hand-written, and that is a normal way to use it.
- 03
Preview the whole set.
Read it back with its scoring and branching laid out, and fix what reads wrong before anyone gets a link.
- 04
Publish when ready.
Then invite a client by name and email. They get their own login and answer one question at a time, with progress saved.
Over time
The drafts get closer to your voice
What you keep and what you dismiss feeds the next set, so the drafts drift toward how you write. Keep dismissing the rating-scale questions and they stop appearing; keep the ones that ask about a specific occasion and more of those come back.
That is the only sense in which the system learns from you. It is a preference signal - which suggestions you kept - held privately in your account. No model is trained on your content, and nothing you write is used to draft for anyone else.
Coverage
Which fields ship configured
HR, finance and healthcare and wellness ship with five specializations each.
HR
Recruitment, Employee Relations, Performance Management, HR Strategy, Compensation & Benefits
Finance
Tax Planning, Investment Advisory, Financial Analysis, Wealth Management, Risk Management
Healthcare / Wellness
Nutrition, Mental Wellness, Physical Therapy, Lifestyle Coaching, Stress Management
That is what ships, not the whole list. More are added over time, and if the specialization you need is not there you request it from your profile and an admin reviews it. Leadership coaches, career coaches and business coaches define their own, which for those fields is the right way round, since no vendor's default list would match how any particular coach works.
Straight answers
What the builder does not do
No template library.
It drafts for your specialization rather than handing you someone else's form. If you want three hundred pre-built questionnaires, that is a different kind of tool.
No validated instruments.
You cannot build a PHQ-9, a DISC or a Hogan here, and you should not try. License those separately and run your own diagnostic alongside.
No public or anonymous links.
Assessments go to clients you have invited. There is no quiz-funnel mode.
No emotion detection on video.
Transcription only, as above.
Nothing added without you.
The AI drafts on request and never edits, inserts or regenerates on its own.
The other half
From builder to report
The assessment is half the product.
When a client submits, their answers become a draft report - executive summary, key findings, strength areas, improvement areas, recommendations - and it lands in your review queue, not their inbox. You rework it, add your note, and approve it.
That half is described on the features page, and the two halves are built to hand off to each other: the scoring and branching you set here are what the report is drafted from.
What is in the draft
- 01
Executive summary
- 02
Key findings
- 03
Strength areas
- 04
Improvement areas
- 05
Recommendations
Questions
Common questions
Last updated . Reviewed by Ankush Tarika, Marketing Head.
One specialization is enough to judge it
Name your field and read the draft that comes back.
Every account starts with a 14-day free trial of the whole workspace, AI included, no card required. Name one specialization, read the draft that comes back, and decide from there.