The Scorvia

• 

  Logo

Inspiration

Most coaches ran assessments the same way: a questionnaire built in Google Forms or Word and adapted from the last client, answers scored by hand, and an evening spent writing each report. The thinking was not the hard part. Turning notes into something a client could read took an hour per person, every time.

The tools that existed either handed coaches a library of someone else's templates or gave them a blank form. Neither started from the question that matters: who is this assessment for?

What it does

Scorvia is an AI assessment workspace for coaches and trainers. You name the field you work in and the AI drafts questions in that field's language. You edit, score, branch and reorder the draft, add your own questions, and publish. Clients answer through their own login, one question at a time, by text or by video. When they submit, the report is already written: summary, key findings, strengths, areas to work on, recommendations. It lands in the coach's review queue, not the client's inbox. The coach edits it, approves it, and sends it.

Sessions, packages, courses, client groups and feedback run in the same workspace. Scorvia ships hosted, white-label, or with source code, and is built by Fly IT Solution.

How we built it

Four decisions shaped the product more than any technology choice.

• Specialization first. Nothing is drafted until the coach names their field. That input is the instruction the model works from, and it is why the questions come back specific rather than generic. • Human approval as architecture. Reports are drafted into a queue and cannot reach a client until the coach has read them. This was decided before report generation was built. • A preference signal, not training. What a coach keeps and dismisses shapes their own future drafts, privately. No model is trained on any coach's content. • Video as transcription only. Video answers are transcribed and the words feed the report. Nothing analyses faces, voices or emotional state. Challenges we ran into • Generic drafts. Early drafts sounded like they could have been written for anyone. The fix was refusing to draft until the coach had named a specialization precisely enough to write towards. • Saying no. Coaches asked for 360-degree feedback, public quiz links, a template library and validated psychometrics. Each is a different product. We chose not to build them and to say so on the website. • Any industry without a fixed list. Shipping HR, finance and wellness configured was the easy part. Letting a coach in any other field define their own specialization and still get a specific draft was harder. • Video on wellbeing work. The obvious build was emotion inference. We left it out. Accomplishments that we're proud of • Specific drafts. Name a precise specialization and the questions come back in that field's vocabulary. • Nothing sends unread. Every report a client receives has been approved by the coach first. • Three industries configured, five specializations each, with custom fields for everything else. • One workspace. Assessments, reports, sessions, courses and groups in one place, with nothing for clients to install. • Stated limits. What the product does not do is written on the same page as what it does. What we learned • The setting is the prompt. The quality of an AI draft is mostly decided before generation, by how precisely the user described what they wanted. • Coaches trust what they can override. A draft they can edit down to a word gets adopted. A result they cannot touch does not. • Stated limits build more trust than claims. • People say more out loud than they type. That is the whole reason video answers exist.

What's next for Scorvia

More shipped specializations, added as coaches request them from inside the product.

Built With

• node.js
• react