Where Vekuri goes from here.

Evidence review

The queue where an administrator reads the evidence an employee submitted and approves or rejects it. Approval fulfils the requirement it targets; a rejection carries a mandatory reason, and the employee hears about either outcome.

Placement and assignments

Put an employee straight onto a class roster, or make a course or a certification mandatory for a department, a job position or one person — without waiting for anyone to enroll themselves. Placement respects the session's capacity, with no override path.

Audit and exports

The append-only trail behind every certificate — fulfilments, attendance captures, evidence recordings, issuance, renewals — exported per employee, per certification or per organization. The trail is the artifact an outside auditor actually asks for.

Organizations

Multi-organization administration: create, configure and suspend organizations, each created together with its first administrator. It is the only surface that crosses an organization boundary, and it audits itself.

Voice practice agent

An employee rehearses the real exam by talking it through, out loud, with an agent that behaves as the examiner for that subject — instead of clicking through multiple choice.

• The conversation runs in the browser, phone included, with no app to install.
• When the practice ends the agent evaluates the attempt and says where it went wrong — criterion by criterion, not as a single number — and points each failed criterion at the lessons to review.
• The guidance reaches past the course at hand into a recommendation across the whole catalog the employee can see.
• A readiness signal read from the attempt history — advisory, labelled that way in the interface, and never a grade.
• It is launched from Practice exam in the employee app, and from the course or certification the practice is attached to.
• Every attempt is kept, so the employee can read it back and practise again.
• Practice stays practice: an attempt never gates, grants or contributes to a certificate.

AI Content Builder

An organization makes its own courses out of the material it already has, with AI drafting the first version. A person always reviews and approves before anything is published — the review gate is part of the design, not a safeguard bolted on afterwards.

AI navigation companion

A companion the employee talks to in order to find their way around: which courses are open to them, how far along they are, what state each of their certifications is in. It answers from that employee's own data and nothing else.

HR bot by email

The bot has an email address of its own, one per organization. Anything that lands there becomes an approval request in the HR dashboard — where the administrator decides, rather than in an inbox.

• An employee whose phone died and who could not check in simply writes an email.
• The message is parsed into a structured intent with a confidence score attached, and filed in the needs-action inbox as an approval request.
• The administrator approves or rejects it there. Nothing executes before a human says so.
• Its write scope is scheduling only — assignments, enrollments and placements, capacity respected, no override. It never writes attendance, fulfilments or certificates.
• An “I was in the room and was not counted” claim is routed to the facilitator to verify, never granted by the bot.
• Every write the bot originates is attributed to a machine actor in the audit trail, beside the human who approved it.