Session log — Discovery AI medical school track analysis

28 May 2026 · Hasmukh with Claude · auto-published from the local journal entry. A polished narrative version can be requested in any future Claude session.

Summary

Hasmukh asked whether the Discovery AI plugin on tibaai.org could be adapted for a medical school attached to a teaching hospital, covering clinical care, cutting-edge research, and the training of scientists and healthcare workers. He felt the existing plugin was geared toward his s2l.online style of need (single-owner commercial sites), and wanted a version that fits the academic medical centre sector.

Read the Discovery AI plugin source on the s2l VPS at tibaai's plugins folder. Confirmed the plugin's actual shape: a guided interview, run in fifteen to twenty minute sessions, that builds up a full website specification and then generates a working homepage in HTML for live preview.

Mapped out the three existing tracks the plugin already supports:

Track A — Digital Products, five sessions, long-form sales-letter homepage with external checkout
Track B — Services and Booking, three sessions, trust-led homepage with a booking CTA
Track C — Something Else, routes the visitor into A or B

Identified why neither A nor B fits a medical school plus teaching hospital. The sector has three simultaneous missions (care, research, education), many distinct audiences, no checkout, much richer trust signals, and a portal-style information architecture rather than a single landing page.

Drafted a proposal for a fourth track inside the same plugin, Track D for Academic Medical Centre. Four interview sessions covering mission and identity, the three estates (care, research, education), audience journeys for seven typical visitor types, and brand voice plus operations. Also sketched a different homepage shape with a three-pillar Care, Research, Education strip near the top, impact numbers, featured breakthrough, student and alumni faces, partner and accreditation logos, news and events, and a multi-CTA footer panel covering Apply, Refer a Patient, Find a Clinician, Partner, Support and Careers.

Presented this back to Hasmukh in plain English and offered two next-step options: draft the actual Track D system-prompt content for review, or produce a sample homepage preview for a fictional medical school so he can react to the shape first.

Decisions

Adding a fourth Discovery AI track for academic medical centres is the right approach, rather than trying to stretch Track A or B.
The new track must be built around the three-mission triangle (care, research, education), not around a single product or single practitioner.
Existing tracks should be left untouched. The work would slot in alongside them, mirroring how Track B was added on top of Track A.
No changes to tibaai.org or any other site were made or proposed yet. The proposal is sitting with Hasmukh for direction.

Changes made

No changes on any server. No content, settings, plugin files or database rows were touched on tibaai.org, medilearn, medistage or anywhere else.
New file written locally only: this session log.

Follow-ups

Hasmukh to choose the next step: either commission the full Track D system-prompt draft for review, or first ask for a one-off sample homepage preview for a fictional medical school to test the shape.
Once direction is confirmed, the Track D work would be a single drop-in extension of the discovery-ai plugin: opening track-picker gets a fourth option, the system prompt gets a Track D section with four interview phases, and the homepage generation prompt gets a three-mission template.
Worth flagging to Kenn at some point that this is being scoped, so he is aware before any work lands on tibaai.

Update — end of day, 28 May 2026

Built the sample preview for Hasmukh to react to. Fictional institution called Stellenberg School of Medicine, saved as track-d-sample-academic-medical-centre.html in the project's previews folder and also copied to the top of the iCloud 44 CLAUDE folder for overnight review. The preview includes the utility bar, three-pillar Care/Research/Education strip, impact numbers, Dean narrative, featured research and clinical cards, voices, partners, news, six-tile audience CTA grid and institutional footer.

Hasmukh then refined the brief. The actual use case is not a whole faculty or school. It is a single department within a Faculty of Health Sciences, specifically the Department of Internal Medicine, the largest department in such a faculty. It has many sub-divisions including cardiology, nephrology, dermatology, rheumatology, endocrinology, neurology, and others. Still sits under an academic medical school attached to a teaching hospital, but the shape of the site is more granular.

Implications for the next session

The Stellenberg preview is the right "whole faculty" reference, but the next sample needs to be department-scoped. Key differences for a department of internal medicine site:

Sub-divisions become the centrepiece grid on the homepage, not the three-mission triangle.
Audience narrows to referring doctors, registrars seeking FCP training, fellows, medical students on rotation, and patients with a specific condition.
Impact numbers shift to admissions, specialist clinics held, registrars trained, FCP pass rates, active clinical trials, recent publications per division.
Academic rituals (Grand Rounds, mortality and morbidity meetings, journal clubs) belong on the homepage.
CTAs become Refer a Patient by sub-specialty, Apply for Registrar Training, Find a Specialist, Join a Trial, Contact a Division Head.
Branding sits under a parent lockup (Department, Faculty of Health Sciences, University, Hospital), not standalone.

Plan for the morning

Build a fresh sample preview for a fictional Department of Internal Medicine, with the sub-division grid as the lead element. Keep the Stellenberg preview as a sibling reference for "whole faculty" scale.
After Hasmukh reacts to the department preview, draft the Track D system-prompt content as a single flexible track that asks at the start of the interview whether the site is for a faculty, a department, or a sub-division, and adapts the interview phases and the homepage template accordingly.
Read both preview files together before drafting the prompt so the prompt captures the differences between scales.