You are one person. When you are asleep and my board portal is down at 2am the night before an AGM, who answers the phone?

The full executive answer

Nobody answers a phone at 2am today, and I will not pretend otherwise — so let me tell you what is true instead, because the honest architecture answer is stronger than a fake staffing answer. The platform is built so that no human needs to be awake for it to stay up or come back: it runs serverless on Vercel with managed, replicated Postgres — there is no server for me to restart at 2am. Recovery targets are published: four hours to restore service, twenty-four hours maximum data loss under daily backups, with point-in-time recovery narrowing that. Monitoring — Sentry, PostHog, OpenTelemetry — pages me automatically; a 2am incident does not wait for a customer email to be noticed.

For the AGM scenario specifically, the design principle is that your critical artefacts are never hostage to our uptime: board reports and packs export to PDF, so the material for tomorrow's AGM should already be in the room. The public status page covers seven subsystems with incident history, so your team sees what we see, and our incident process runs a full disclosed lifecycle — open, investigating, contained, resolved, post-review, closed — with the post-review published.

Now the roadmap, plainly: founding and enterprise customers today get a named contact — me — with a four-business-hour first-response commitment, and founding customers have my personal number. A formal L1/L2/L3 support organisation with true 24×7 coverage is a funded roadmap item that arrives with revenue, and I will put that milestone in writing in an enterprise order form. If genuine 24×7 human coverage is a day-one hard requirement, I would honestly tell you to wait two quarters — and I suspect the honesty of that answer is worth something in a vendor.