InterviewReady

Practise for a specific job, see evidence-linked feedback against a frozen rubric, and improve the next attempt.

Comment

Business Viability

Job seekers can read advice, but improvement requires speaking under pressure and understanding whether the next attempt is actually better. Generic voice chat provides conversation without a stable role-specific rubric, evidence-linked feedback, or a trustworthy attempt history. InterviewReady turns the job description into the system of record: the candidate confirms a generated rubric, answers turn-based questions by voice or text, receives cited feedback, and can compare retries against the same frozen criteria.

The initial buyer is a UK job seeker with a confirmed interview or active shortlist. The current offer includes one first attempt and a £19 monthly plan for continued practice. The product will test monthly access against session packs because interview preparation is episodic. Acquisition begins with role-specific search pages, short demonstrations, careers communities, and referrals from coaches or employability programmes. Activation is a completed three-answer practice loop with feedback viewed. The strongest retention signal is a second attempt against the same rubric, not passive content consumption.

Production includes self-serve accounts, persistent rubric and attempt history, entitlement consumption, a live Stripe Checkout path, and audit logs. The evidence pack currently contains no verified independent product revenue or paying customers, so the business case is presented as a tested operating product with an unproven acquisition model—not as invented traction. The next milestones are observed practice sessions, usefulness ratings, retry rate, paid conversion, refunds, and measured model cost per completed attempt.

The official cash-basis P&L records $0 revenue and $0 product-specific cash expenses. Founder time and shared Yensi hosting, subscriptions, and product-factory resources existed before this entry; the submission discloses them without converting shared resources into InterviewReady-specific cash spend.

We started InterviewReady on 3 August 2026, only two weeks before the deadline. In that compressed period we chose to prove the end-to-end practice, persistence, audit, and safety architecture rather than simulate market maturity. The honest commercial baseline is £0 revenue and 0 paying customers.

AI-Native Operations

Gemini drives two core decisions. First, it converts a real job description into a structured, job-specific rubric that the candidate reviews and freezes. Second, it extracts evidence from stored answer turns and proposes criterion-level coaching signals. Production logs show successful rubric generation and answer scoring with request identifiers, model name, input/output token counts, latency, timestamps, user, rubric, and attempt linkage.

The product—not the model—owns authority. Confirmed rubrics are versioned and immutable. Every committed criterion needs a quote from the stored transcript; missing evidence can remain insufficient. Deterministic code owns entitlement, question order, attempt state, rubric version comparisons, score bounds, history, and deletion. Gemini cannot silently alter a transcript, infer protected traits, invent candidate experience, predict a hiring decision, or grant access. Raw audio has a bounded retention policy while transcripts and audit evidence persist.

This is operational AI rather than generated advice pasted onto a page. The model output changes the rubric and evidence map that power the practice loop, while versioning and citations let the candidate inspect the basis of the feedback. If Gemini is unavailable, the system records the fallback rather than labelling it as a successful model call. Automated verification covers signup, job-description intake, rubric freeze, three-answer flow, evidence-linked scoring, entitlement, persistence, AI usage, and checkout. The current backend suite reports 60 passed with one optional integration skip, and the production evidence records successful live Gemini operations.

Category Impact

InterviewReady targets job access, preparation confidence, and repeatable learning. The theory of change is that role-specific practice produces more relevant questions; evidence-linked feedback turns vague encouragement into an inspectable improvement target; a frozen rubric makes retries comparable; and preserved history helps the candidate see progress instead of starting over for every session.

Impact will be measured through rubric confirmation, practice completion, transcript correction, feedback views, retry within seven days, score/evidence change on the same rubric, user-reported usefulness, and paid continuation. The product will not claim that it causes job offers. Scores are coaching signals, never employer predictions, and the submission explicitly avoids employment-outcome causation.

The same system can later support reviewed role-family packs and user-controlled coach reports without turning into an employer screening product. Shared Yensi product-factory contracts and runtime rails existed before the competition; the InterviewReady workflow, UI, domain logic, persistence, Gemini operations, and billing path were built inside the eligible window.

Built With

Share this project:

Updates