Our editorial approach

Show the work behind the product.

Case studies, tool recommendations, and building guides should make clear what was built, what is known, and what still needs checking.

Who is responsible

Pretectal is the product-building practice of Eugene Huang, O.D. The case studies and guides share his work and approach to building useful products with AI. A byline identifies responsibility; it does not imply independent review.

Editorial contact: eugene@pretectal.com. Founder profile ↗

Current production and review status

The workflow guides, product stories, tool guide, starter resources, review worksheet, and BLINK example were drafted and implemented with AI assistance, including writing and code generation. Source links and the example's published values were checked during production on October 2, 2026.

The specialist healthcare resources have not undergone independent clinical peer review, CE accreditation review, or a study of learning effectiveness. We do not claim that a completed worksheet establishes accuracy, approval, or accreditation. Our workflow recommendations describe the review a production team should perform.

How we report product evidence

WhatsTheDx's 1,100+ sign-ups and the description of paying students using OptoConnections daily were supplied by Pretectal's founder on October 2, 2026. The sign-up figure is cumulative; the OptoConnections description is qualitative. Neither is a live analytics feed or an independently audited measure.

Product features are described from the public experience and implementation. OptoLearn is presented as a working training platform, without a claimed adoption count or reduction in onboarding time. Product screenshots are dated captures and may differ from the current interface.

Adoption, payment, and repeat use are distinct from learning outcomes. Our product stories do not establish improved clinical performance, exam results, or workplace readiness. Suggested prototype exercises are separate from the reported development history of the existing products.

How we describe AI development and tools

Eugene reports starting the MVPs with Lovable, Cursor, and Windsurf, then moving to Claude Code and ChatGPT Astra with skills. This describes the shared evolution of his workflow. We do not assign a particular tool or framework to an individual product without a project-specific record.

Tool capabilities are linked to official documentation. Our recommendations explain which task a tool may suit; they are not comparative benchmarks or promises of time saved. Tool features and access can change. The AI tools guide distinguishes the development assistant, application framework, backend, and hosting decisions.

Read the tools and stack guide →

Keep different kinds of material distinct

  • Published findings: identify the original source, relevant population, endpoint, and limitations. Preserve reported contrasts when rounded group values differ.
  • Teaching models: explain meaningful simplifications and avoid presenting the model as a measurement or individual prediction.
  • Fictional cases and synthetic data: label them where learners encounter them.
  • Decorative artwork: distinguish it from clinical evidence. The homepage iris is generated artwork, not a patient photograph; its pupil animation is illustrative.
  • Production guidance: distinguish our recommendations from an accreditor's requirements and a client's own procedures.

Make review specific

Record the version, task, responsible reviewer, issues found, and changes made. Bring in appropriate clinical or illustration expertise when the material exceeds the team's competence. AI output can help organize a review, but agreement between generated answers does not establish validity.

Where an activity has organizer or accredited-provider requirements, use that organization's review and approval process. Company promotional content and independent accredited education require distinct planning and ownership.

Respect sources and collaborators

Link to original sources and distinguish a citation from permission to reproduce an image or other protected material. Record licenses and permissions for reused assets. Attribute the people who contribute scientific judgment, illustration, and instructional work.

Corrections and updates

Send the page URL, the disputed statement or image, and a supporting source to eugene@pretectal.com. We assess reported issues and update the affected resource when a correction is warranted. Dates change when the content is substantively revised.

Initial workflow collection: October 2, 2026. The guides are curated resources; they do not automatically refresh when a source or provider policy changes.

Sources and further reading

Source links checked October 2, 2026. Our production workflows are editorial recommendations; provider requirements apply to their own activities.

Responsible use of AI in accredited continuing education (opens in a new tab) ↗

ACCME · January 30, 2026

Guidance on disclosure, oversight, and independence in ACCME-accredited education.

Content validity for AI-enabled clinical learning (opens in a new tab) ↗

ACCME · September 22, 2026

Distinguishes reviewed educational materials from dynamically generated clinical answers.

Standards for Integrity and Independence (opens in a new tab) ↗

ARBO / COPE

COPE's adopted standards for content validity and independence. Check the applicable provider's requirements.