Great breakdown. Did you notice any memory differences under high concurrent load with Turbopack?
This resonates a lot. Outreach quality always beats raw volume in B2B SaaS.
The proxy.ts convention in Next.js 16 is much cleaner than the old middleware approach once you get used to it.
Returning structured results and handling redirects client-side eliminates any NEXT_REDIRECT catches in modals.
Are you seeing any caching friction with server actions inside dynamic modal forms?
How did you structure your early feedback surveys without creating onboarding friction?
Optimistic local state for immediate interaction feedback combined with Supabase Postgres Changes subscription gives the best of both worlds.