Change Log
UI Fixes — Interview Flow
The interview agent’s question logic has been rewritten. Instead of asking the same TRIPS framework questions verbatim for every pain point, the agent now varies its phrasing and order conversationally, and uses natural bridges between topics — for example, “That makes sense — is there another area that drains your time?” The interview will feel much more like a natural, dynamic conversation rather than a scripted questionnaire.
Silence detection was already implemented with well-calibrated thresholds: 15 seconds of inactivity triggers a “Still there?” prompt, 25 seconds triggers an auto-end, and a 2-minute minimum conversation age prevents false positives at the very start of a session. No changes were required here.
Logic has been added to detect when the agent announces a transition to a new topic during speech. When transition phrases are recognized — such as “let’s move on,” “next section,” or “shifting to” — the on-screen status updates in sync with the voice, so the UI no longer lags behind what the agent is saying.
The “End Segment” button now has two requirements before it appears: (1) at least 3 minutes of elapsed time, and (2) the agent must have spoken at least once. In addition, clicking the button now presents a confirmation dialog — “Are you sure you want to end this segment?” — before taking any action, preventing accidental early termination.
A retry counter has been added per interview segment. If a segment is scored incomplete (“needs more detail”) multiple times, the system now offers a “Continue Anyway” option after 3 attempts rather than looping indefinitely or terminating the interview prematurely. The interview completion flow has also been corrected to always display the closing screen once a session is marked complete — regardless of individual segment scores.
Landing Page
User name was already stored in the system from the interview invitation. A new Title / Role field has been added to the user profile data model (e.g., VP of Operations). Title is now captured at the point a user account is created in the admin panel and is displayed on the interview landing page alongside their name — for example, “Welcome, John Smith / VP of Operations, Acme Corp.”
Organization name was already stored in the system but was displaying incorrectly — a bug in the landing page was reading the user’s name field instead of the company field, causing the user’s name to appear twice. This has been corrected. The right organization name now appears on the landing page.
All four privacy statements were already present on the landing page: Zero-Training Guarantee, Private Encryption, Data Ownership, and Confidentiality. No changes were required.
Intro Screen
The “Your Next Step in Strategic AI” headline and the full Solomon Engine body copy — “Congratulations on making it to this point...” — were already present. No changes were required.
All five preparation items were already in place: A Quiet Space, Your Top 3 Friction Points, Strategic Goals, Physical Context & Focus, and Honest Feedback. No changes were required.
The intro screen content has been reformatted for readability: text is now left-aligned, broken into clear paragraphs, and key phrases are bolded for scannability. The content is unchanged — only the presentation has been refined.
Closing Screen
The closing screen previously showed only a generic list of segment names (e.g., “Pain Points & Workflow”). It now displays a personalized summary for each segment — a brief description of the key insights captured during that portion of the interview, drawn directly from the AI’s completeness analysis.
The closing screen previously displayed a generic “The SuperAssistant team will analyze your interview.” It now shows a configurable advisor name — for example, “Your dedicated advisor, Aparna, will review your responses and be in touch.” This can be customized per deployment without a code change.
“You’ll hear from us within 2 business days” was already explicitly stated on the closing screen. No changes were required.
A prominent “Thank You!” heading with a confirmation message was already present on the closing screen. No changes were required.
A new 1–5 star rating widget has been added to the closing screen, along with an optional text feedback field. Ratings are stored in the database for review. After a user submits their rating, a feedback and contact link appears below — creating a clean, low-friction feedback loop at the end of every interview.
Testing & Verification
All changes have been tested and confirmed working in production at app.thebradburygroup.net.
| Test Suite | Result | Notes |
|---|---|---|
| Backend unit tests | 69 passed | All API routes, models, and services |
| Frontend unit tests | 99 passed | 3 skipped (pre-existing, unrelated) |
| End-to-end tests (Playwright, live site) | 130 passed | 1 failure ∗ — pre-existing, unrelated to this release |
∗ The single failure (STATS-02) is a pre-existing test infrastructure issue — a browser security restriction (CORS) prevents that particular test from calling the backend API directly during a Playwright run. It was failing before this release and is not related to any of the changes above.
SuperAssistant — All changes are live. New interview sessions will reflect all updates going forward.