Mental Edge

Mental Edge had sentiment analysis as capability. The business needed a product: a portal coaches and teams could log into, trust, and pay for. I focused on the product shell — UX for insights, auth, Stripe, and the paths that make AI output usable in a daily workflow.

Coaches and teams needed emotional/sentiment insight from unstructured text — not a research notebook, but a repeatable product workflow they could buy.

Accuracy alone doesn’t create revenue. Without auth, billing, and a clear insight UX, the model stayed a demo instead of a subscription business.

Raw analysis without product structure meant no account model, no purchase path, and no reason for non-technical users to return. Capability without packaging doesn’t close deals.

Coaches and teams reviewing text-based signals; operators who need secure access and a billing path that supports GTM.

• –Model output had to become decisions people trust — not opaque scores
• –SaaS shell (auth + billing) had to support go-to-market
• –AWS hosting patterns that won’t surprise ops later
• –Analysis latency acceptable for daily use

Frontend + product engineering — portal UX, auth, Stripe-ready SaaS shell

React portal talking to a Django REST API on PostgreSQL, hosted on AWS. Kinde for auth, Stripe for billing. The product surface prioritizes insight workflows over exposing model internals.

• Productize via authenticated portal + Stripe

  Why: GTM needs accounts and a purchase path before more model tuning.

  Rejected: Shipping more model accuracy first — doesn’t help if nobody can buy or return to a workflow.

• Kinde for auth

  Why: Faster, safer auth than building identity from scratch for an early SaaS shell.

  Rejected: Custom auth early — high risk and distraction when the product still needed UX clarity.

• Keep insight UX opinionated

  Why: Coaches need decisions, not dashboards of raw model dumps.

  Rejected: Exposing every score and parameter — impressive to engineers, confusing to buyers.

• –Turning model output into decisions coaches trust
• –SaaS shell (auth + billing) that supports GTM
• –AWS hosting patterns that won’t surprise ops later
• –Keeping analysis snappy for daily use

• –Speed to a billable portal vs. deeper model work — we chose the path that unlocks revenue learning.
• –Managed auth vs. full control — Kinde reduced build time; identity edge cases become vendor constraints.
• –Opinionated insight UX vs. power-user flexibility — clearer for coaches, less room for analysts who want raw dumps.

Built and shipped the portal experience around insights, wired auth and Stripe, and connected the Django/Postgres/AWS backend so the AI capability lived inside a product people can actually use.

Moved from “show the model” to “sell the workflow” — packaging, access, and billing became first-class, not afterthoughts.

I’d define success metrics for the insight UX earlier (what a coach does after seeing a result) and instrument those paths before adding more model surface area.

Mental Edge could sell access, not just accuracy — a portal with a subscription path instead of a one-off analysis script.

• Paid portal with Stripe
• Secured user access (Kinde)
• AWS production environment
• Text → insight workflow teams can repeat

• –AI products fail at packaging more often than at models.
• –Auth and billing are product features when you’re selling SaaS.
• –If a non-technical user can’t act on the output, the model isn’t done.

• React
• Redux
• Tailwind
• Kinde
• Django REST
• PostgreSQL
• AWS
• Stripe