Yesterday we shipped Private Brain Phase 1.5 and celebrated. Today we learned Ragie.ai is sunsetting. That's the company powering our knowledge management layer. They're going away. Not in 6 months. Soon.
That's an existential threat to the product we just shipped. Customers uploading documents, using Private Brain features, relying on search — all of it dependent on Ragie infrastructure. If Ragie shuts down tomorrow, Private Brain stops working. That's unacceptable for a production system.
So we did what you do in a crisis: we moved fast. 2-day sprint. Multiple deployment attempts. Environment variable hell. Git workflow learning curve. GitHub setup for a team unfamiliar with version control. But we got there. We now own the entire Private Brain infrastructure. Ragie was a foundation we borrowed; now we've built our own.
Ragie.ai Sunset: The vendor powering Private Brain knowledge management is sunsetting. This creates immediate vendor risk for all customers currently using or planning to use the platform. Private Brain features are now dependent on external infrastructure with unknown availability timeline.
SENDGRID_API_KEY, SUPABASE_ANON_KEY, SUPABASE_URL. Multiple deployment attempts with mismatches resolved. All secrets properly injected into deployment pipeline.Netlify deployment pipeline failed on first attempt. Build succeeded, but runtime failed. Application tried to connect to Supabase — credentials missing or malformed.
Root cause: Environment variables not injected into deployment environment. SUPABASE_ANON_KEY undefined. SUPABASE_URL undefined. Sendgrid key missing. Application crashed on startup.
Added all required environment variables to Netlify build settings. Mapped each secret from local development environment to Netlify's encrypted secret management. Verified each variable resolves at build time. Re-deployed.
Second issue: Environment variable names changed between repos. Local dev used REACT_APP_SUPABASE_URL. Netlify environment used SUPABASE_URL. Mismatch. Fixed all references. Deployed again. Success.
Landing pages loaded but images didn't. Blank spaces where logos should be. Broken image errors in browser console.
Root cause: Filename inconsistencies. Asset files named Hyvara_Logo.png but HTML referenced Hyvara Logo.png (with space). Asset files named with underscores but HTML referenced with dashes. Case sensitivity mismatches.
Audited all image references across landing pages. Corrected all file path references to match actual filenames exactly. Changed all spaces to underscores in asset filenames for consistency. Updated HTML <img src=""> tags to reference correct paths.
Added asset versioning to prevent caching issues. Verified all images load in development and production. Tested across different browser tools.
Landing page forms submitted but users weren't redirected. Test drive users should go to app. Early access users should go to confirmation page. Neither was happening. Form stuck. GA4 events not firing. Conversion tracking broken.
Root cause: Multiple issues. Form submission handler didn't exist. Redirect logic not wired. Event tracking code not hooked to submission. Cascading failures.
Built form submission handlers for both landing pages. Wired conditional logic: if test drive, redirect to app URL; if early access, redirect to confirmation. Added GA4 event firing on form submit. Tested both paths end-to-end.
Verified Google Ads conversion pixel fires on submission. Confirmed attribution chain works. Users now convert correctly. Funnels flowing.
Team unfamiliar with Git and GitHub. Multiple people making code changes without proper version control. Merge conflicts. Lost work. Deployment confusion. Who pushed what? When? Why?
Under crisis mode, this friction could kill productivity. Needed to establish Git discipline fast.
Established GitHub repository with standard folder structure. Configured Netlify for continuous deployment from main branch. Set up branch protection: all changes require pull request review before merge to main.
Documented workflow: create feature branch → make changes → commit with descriptive message → push to GitHub → create PR → get review → merge to main → Netlify auto-deploys.
Trained team on cycle. Now everyone can independently contribute code without stepping on each other. Deployment is automatic. No manual steps.
For Adobe: When Mike's dedicated person says "Hyvara owns all its infrastructure," that's a differentiator vs competitors who bolt together SaaS tools. You own the stack. You control the uptime. You can customize without permission.
For enterprise buyers: Vendor risk is always a concern. Ragie's sunset could have killed us. It didn't, because we moved. That's proof we take risk seriously. We own infrastructure. We're built for scale.
For the team: We just lived through a crisis and came out stronger. That builds confidence. We can ship features fast because deployment is automated. We can fix bugs immediately because diagnostics are transparent. Infrastructure ownership is power.
Infrastructure ownership is defensible. Competitors who use Ragie or similar vendors are now vulnerable. We own ours. As we scale, infrastructure becomes harder to replicate. That's moat.
Custom optimization is now possible. We can build features that depend on infrastructure details. Vendors can't do that. They're constrained by the platform they're building on. We're not.
The Ragie crisis proved our resilience. Mike Neumann and Adobe are watching. When August planning happens and someone says "But what if your vendor goes down?" We can point to the last 48 hours. We moved. We own it now. Crisis proven resilience is worth more than any sales pitch.