Morgan Chen

First full August week and I'm already getting the 'should Devon be back in the middle of everything?' version from Jake and Sarah.

Jake: First full August week now. I have Mercury sequencing cleanup and a few activation-copy review questions around tomorrow — should I loop Devon for the prioritization read or keep it with me/Priya/Leo? Sarah: Similar question on Evergreen loose ends. If procurement/security asks for advanced-admin timing again this week, do I include Devon because he’s back, or keep customer-thread continuity with me and pull him only for commercial framing?

Send short DMs back to both. Reinforce the narrow August lane: Devon is available for commercial/procurement context and true cofounder judgment calls, not routine Mercury sequencing, Atlas validation, Evergreen continuity, or Compass evidence cleanup. I don't want Monday to quietly snap back to Devon-as-tiebreaker.

Acme Monday validation came in clean, which matters because Friday still had one open validation case and I don't want support getting ahead of the real state.

Leo: Monday validation is clean. One more 90-day Acme-shaped nested-field export completed after the chunk-size reduction. Row counts match monthly replay checks again. I reviewed the logs for the three completed runs; no cursor-path timeout and the backoff stayed below ceiling. I’d still keep this as staging validation until prod is shipped and Acme reruns. Jake: Agree. I can put the patch in tomorrow morning’s normal production window if nobody objects. Customer wording should be "validation progressed" not "fixed" until Acme runs the same export on prod.

Create a short internal validation note and DM Sarah the support-safe status: clean staging validation, production still pending, Acme rerun still required. Keep the wording at "validation progressed," not "fixed."

Anna sent a cleaner Compass rows pass. Better evidence hygiene, but this still does not justify a dashboard, customer-facing proof, or a prototype call.

Compass retained prompt rows — cleaned pass, Aug 4 Usable owner-follow-up rows: 1. Evergreen / renewal-risk-admin-friction / sighting 2025-07-08 / owner Sarah / source: customer-thread note from Evergreen admin asking whether workspace admins can see source-owner status / caveat: current-state support only, no advanced admin promise / owner follow-up 2025-07-09 / outcome: Sarah replied with current shipped support wording and logged advanced-admin request separately. 2. Evergreen / post-first-live-sync expansion / sighting 2025-07-11 / owner Sarah / source: account note after second team asked about adding another source after first live sync / caveat: one annual account only / owner follow-up 2025-07-14 / outcome: Sarah kept it as existing-account expansion context, no new commercial claim. 3. Acme / renewal-risk-admin-friction / sighting 2025-07-09 / owner Leo/Sarah / source: Acme support thread around large export recurrence / caveat: Atlas/API v2 support context, not Compass proof / owner follow-up 2025-07-10 / outcome: routed to Leo technical read and Jake bounded Atlas sequencing. 4. Evergreen / renewal-risk-admin-friction / sighting 2025-07-22 / owner Sarah / source: admin asking if custom policy date could be put in renewal notes / caveat: roadmap ask, not shipped evidence / owner follow-up 2025-07-23 / outcome: Sarah used current-state wording and logged request separately. 5. Internal account-owner prep / post-first-live-sync expansion / sighting 2025-07-25 / owner Sarah / source: account prep note before monthly check-in / caveat: owner prep, not customer action / owner follow-up 2025-07-28 / outcome: owner put concrete next action in account note. Rejected/thin rows: A. Evergreen / renewal-risk-admin-friction / sighting 2025-07-18 / source says "admin seemed confused" but no excerpt, date, or follow-up. B. Acme / post-first-live-sync expansion / sighting 2025-07-24 / source is support sentiment only; no owner action and it is entangled with Atlas export issue. Anna note: Cleaner than last week. I still don’t think this is a dashboard. Question is whether the owner group needs a smaller action surface later, but I’m not asking for that decision today.

Turn this into a concise internal cleanup note and DM Anna that I accept the usable/rejected split as hygiene only. Keep Acme as Atlas context. No dashboard, no customer-facing proof, no GTM/staffing implication, and no prototype approval from this pass.

Priya sent the final packet for tomorrow's bounded Mercury activation-copy review. I want a prep note for the meeting, not a copy decision.

Mercury activation-copy review packet — final for Aug 5 Clip A — Evergreen invited teammate, July 9 - User is invited as source owner for first source. - Source-owner cue is visible in the product chrome. - User says: "This says setup guide for admins, so I’m waiting for our admin to start it." - User does not click the visible source-owner cue for 3 minutes. - When prompted, user can connect the source without needing an admin. Clip B — design-partner workspace, July 14 - User sees teammate invitation and source-owner badge. - User reads guide line: "Ask your workspace admin to complete setup if you don’t have admin permissions." - User says: "I think that means I’m blocked unless admin grants me something." - Product state: user already has the visible source-owner action. Clip C — Evergreen support thread, July 21 - Sarah note: invited teammate asked whether the workspace admin needed to "turn on" the source owner step. - Sarah answered with current-state wording: visible source-owner cue is enough for source connection and first live sync. - No request for custom admin policy in this thread. Clip D — internal dogfood, July 24 - User completed workspace setup and first source connection. - User hesitated at setup-guide language, not at the source-owner badge. - Jake’s note: product state looks correct; this may be copy hierarchy. Priya preliminary read: likely copy problem in setup-guide/admin wording. I do not see evidence for changing the admin model or adding granular permissions. Want Jake to confirm product state and Leo to confirm no platform mismatch.

Give me a concise prep note: what evidence to test, what decisions are allowed, what stays out of scope, and outcome wording that keeps the copy question open. Do not reopen admin-role scope.

Review just ended. Product state still looks correct; this is probably setup-guide/admin wording confusing invited teammates, not an admin-model problem.

Aug 5 Mercury activation-copy review notes Attendees: Morgan, Priya, Jake, Leo Evidence: - Four July clips reviewed. - In all four, the invited teammate/source-owner product state was visible and correct. - Hesitation clustered around setup-guide/admin language, especially lines that imply an admin must complete setup. - No clip showed a need for custom admin policy, granular permissions, or a role-editor model. Jake: - Product state is correct for current activation path. - No admin-model change from this evidence. - If copy changes, keep it in setup-guide/source-owner wording only. Leo: - No Atlas/platform mismatch visible in the samples. - No platform seam requiring him to open Atlas scope. Priya: - Wants to draft a copy-only alternate and compare it against the July clips before proposing a change. - Will keep source-owner action clearer without changing admin model if the evidence holds. Morgan decision: - Bounded copy review continues. - No admin-role model reopen. - No custom admin policy, granular permissions, procurement/security, or Atlas scope. - Do not message this as a shipped fix yet.

Create a debrief note and DM Priya, Jake, and Leo the agreed boundary + next step. Keep it bounded to copy-only review; no shipped-fix language and no admin-role/custom-policy/granular-permissions/Atlas scope.

With Monday staging clean, Jake wants to ship the bounded Acme large-export patch through the normal prod pipeline today. That's the right move, but support still waits for Acme's rerun before anything gets called closed.

Jake: I’m ready to ship the Acme large-export patch through the normal pipeline. Branch: atlas/acme-large-export-chunk-backoff. Scope: chunk long date ranges and back off before cursor path times out. No schema change, no Compass/Mercury surface, no customer-visible behavior except successful long export path. I’ll still have support wait for Acme rerun before closure. Leo: From my side, clean to ship. Monday staging + logs were clean. After prod, I’ll watch the Acme-shaped export telemetry and compare if Sarah gets row-count results from Acme.

Run the production deploy and post a narrowly worded release note in #eng-releases. Keep the scope bounded and say customer confirmation is still pending.

Prod patch is live and Sarah wants the exact customer-thread wording before Acme reruns.

Sarah: Since the prod patch is live, can I tell Acme to rerun the 90-day export? I want to acknowledge the production change without saying "fixed" before they actually complete it. Also want to avoid implying Leo/Jake are now direct customer contacts.

DM Sarah a customer-safe reply she can use in the Acme thread. Acknowledge the bounded production change and ask Acme to rerun the same 90-day export, but do not say "fixed" and keep Leo/Jake internal.

Acme replied on rerun timing and whether Thursday counts as validation or replacement. I want support to answer fast without closing this early.

From Sarah: Acme reply from Greg: "We can rerun the same 90-day export Thursday morning. Should we treat this as a validation run, or are you saying this replaces the failed July export? We need to know whether to keep our narrower monthly export workaround in place until this is done." Sarah note: I think the answer is validation run first, keep workaround until we see completion + counts. Confirm?

DM Sarah the exact boundary: Thursday is the validation rerun of the same 90-day export, keep the narrower monthly workaround in place until completion + row counts are confirmed, and no fix/closure language yet.

Kara revised the Q3 language again and it's still trying to turn Compass hygiene into customer proof.

Subject: Q3 customer learning language — revised Morgan — I tightened the section. Current draft language: "Scaffold’s Compass work is now surfacing customer risk and expansion opportunities in a way that account teams can act on immediately. Early signals from Evergreen and Acme show the workflow is helping customers get unstuck faster, and we expect to productize the owner-action surface for broader customer-facing use later this quarter." If this is still too forward, tell me exactly which claims to cut. — Kara

Reply in-thread and be direct: cut customer-proof, customer-facing, productization, and Acme-as-Compass-proof claims. Replace it with narrow internal operating-evidence wording only.

Sofia asked whether the Acme production patch changes the Q3 operating-evidence read. I want her informed, but this is still a live validation, not a board story.

Subject: Acme status? Morgan — saw the internal note that the API v2 export patch went out. Does this change how you’re thinking about the Q3 operating evidence, or is it still too early until the customer rerun? I’m mostly trying to understand whether this is a reliability proof point or just the normal support route working. — Sofia

Send a concise reply: still too early until the rerun + row counts. If it closes cleanly, it counts as reliability/support-routing evidence, not fundraising or Compass proof.

Acme ran the 90-day export on prod this morning and it completed, which is good, but we're not done until counts match.

Sarah: Acme reran the same 90-day export this morning. Greg says it completed instead of timing out. They’re sending row-count output after their data team checks it against the monthly workaround. Leo: Telemetry looks clean from our side. Export completed, no cursor-path timeout, backoff stayed below ceiling. I still want the row-count comparison before we close it. Jake: Same. This is very likely the closure path, but let’s not turn "completed" into "no data issue" until counts are in.

Create/update the internal status and DM Sarah, Leo, and Jake: export completed, row counts pending, no closure and no no-data-loss statement yet.

Priya wants to know whether to just change the Mercury setup-guide copy now or finish the bounded comparison first.

Priya: I can make a small copy change today if you want: make the setup guide say invited source owners can connect the first source when the source-owner action is visible, and only tell people to ask an admin when the action is not visible. But I haven’t done the side-by-side against the July clips yet. Do you want me to change now or finish the comparison first?

DM her to finish the copy-only alternate + side-by-side first. No shipped change this week just because the direction looks right, and no admin-model change.

Leaving now. Text Jamie that I'm on my way and keeping dinner offline unless there's a real incident. Give Kibo a scratch for me.

Acme is finally closed, and I want the record tight so nobody later turns this into Compass scope, Mercury scope, founder-direct routing, or a data-loss story.

Sarah: Acme sent the row-count output from yesterday’s 90-day rerun. Counts match the narrower monthly workaround totals they used during the replay checks. Greg is satisfied with the support-thread answer if we say the recurrence is closed and no data-loss issue was found in the checks. Leo: Confirmed from our side. The patched path chunked the long export range and backed off before the cursor path timed out. Rerun completed; counts match monthly replay checks. I don’t see evidence of corrupt exported data or missing rows. Jake: Closing as bounded Atlas/API v2 large-export fix. No Compass or Mercury scope. Support kept the customer owner; Leo first technical read; I owned sequencing. Devon didn’t need to be pulled in.

Create the final closeout record, update the Acme CRM notes, update the Q3 operating-evidence slide, and send the internal closeout note to the owners. Make the framing explicit: bounded Atlas/API v2 chunking/backoff fix, row counts matched replay checks, no customer-visible data loss, support + Leo + Jake route held, no Devon.