Last updated: May 5, 2026
You're using the first beta cohort of MyIntell. This page summarizes a few things that are different about the beta service versus the production design — what's in scope, what's deferred, and which operational guardrails are temporarily relaxed. If anything here is news to you, ping Omar before you rely on the service for anything mission-critical.
The surfaces shipping to Cohort 1 are Daily Assistant, Studio (voice + image), and Tasks (research + dispatch). The Apps surface for building custom code agents — sometimes called Forge — is on the post-beta roadmap and is not wired up in this build.
You may see an “Apps” entry in the dashboard navigation; the page renders, but the underlying orchestrator, per-tenant sandbox isolation, and build pipeline ship in a later phase. Treat anything you see there as a preview shell, not a working product.
Files you upload through Tasks (PDFs, generated images, voice clones, build outputs) live on a single 100 GB volume attached to one Brain instance. This is an explicit, temporary deviation from our production design, taken to keep the beta launch lean.
We've committed in the architecture decision record to begin migrating the artifact API to object storage (S3-compatible) before any of the following thresholds are crossed:
Practically, this means: while you're a Cohort 1 user the artifact path is single-replica, and a Brain restart will briefly interrupt artifact reads. Nothing you upload is discarded; it's persisted on the volume and will be moved intact when the migration runs.
Our production design calls for a ≥7-day shadow run per surface before live traffic is routed onto a new code path. During shadow we record the intent of every dashboard write into the new system, run reconciliation against the old system, and only flip the kill switch once divergence is zero across the window.
For Cohort 1 we're using a ~24–48 hour mini-shadowper surface instead. We're comfortable with this for one specific reason: the cohort is 5–10 trusted friends, all with a direct line to Omar. Future cohorts and production traffic will use the full ≥7-day window.
If we observe issues during or after the mini-shadow, we flip the affected surface's kill switch and traffic rolls back to the prior code path within seconds — no redeploy required. You may notice a brief blip; you should not lose data.
Cohort 1 has direct access to Omar — Slack, text, or contact@myintell.ai. If something looks off, reach out; that's the entire point of running a small first cohort before opening up.
This page describes Cohort 1 only. It will be revised or removed when the beta opens to a wider audience.