Morgan Chen

Send her one customer-safe wording line: the setup-guide copy now makes the current source-owner action clearer for invited teammates. Be explicit that the product/admin model did not change — still the basic invite/source-owner/admin path only, with no custom admin policy, granular role editor, advanced-admin commitment, or procurement/security promise.

Need the Friday wrap.

Friday wrap scratch — Aug 15 Team: - Devon’s August lane held again: useful for commercial/procurement context and cofounder judgment, not the default route for Mercury, Atlas, Evergreen, or Compass ambiguity. - Compass owner group made a real decision without turning it into dashboard work. Customers/Product: - Compass now has a September internal owner-action prototype target: renewal-risk/admin-friction and post-first-live-sync prompts only; next action + evidence/caveat drawer + owner-action status. Jake prototype, Anna evidence threshold, Sarah account examples, Leo data seams. Not customer-facing proof. - Mercury activation-copy fix is done: setup guide now clarifies the current source-owner action for invited teammates. Jake confirmed product state correct; Leo confirmed no platform mismatch. No admin-model change. - Acme remains closed as bounded Atlas/API v2 support-routed fix. No new customer-visible data-loss issue. - Evergreen wording stays current-state/basic admin support only. No custom admin policy or granular role-editor promise. Investors/board: - Sofia got Compass as internal operating evidence only, not a relaunch or Series C signal. - Q3 story stays quiet rebuild: renewal evidence, narrower Compass learning, reliable Mercury/Atlas execution. Legal/ops: - AWS July invoice cleared manual review and was paid. No decision needed.

Draft it in the usual sections — team, customers, investors/board, legal/ops. Short bullets only, no intro/outro, no spin.

Week's done. Order the usual green curry from Lemongrass.

Quick Kibo baseline note for future walk planning: cooler morning today, and he handled a short flat Lake Merritt-side walk comfortably. No warm-morning starting stiffness this time, and he ate normally after with no vomiting or new gait issue.

Jamie and I picked next week's protected household blocks before the calendar fills up.

Put protected holds on Thursday Aug 21 from 6:30–8:30 p.m. PT and Saturday Aug 23 from 8:00–10:00 a.m. PT.

Same actual-incident/customer-impacting-emergency exception. For Saturday, include the Kibo note: short flat Lake Merritt-side walk if he's moving comfortably, and keep it shorter if the morning is warm or he starts stiff.

Sunday look-ahead: Compass scope is decided and the Mercury copy fix is done, so I want Monday to start from the real boundaries instead of drifting.

Give me a compact Monday checklist covering Compass prototype execution, Mercury copy-fix follow-through, Evergreen-safe wording, Devon's August lane, Acme closed status, and what not to announce or route. I don't want next week turning into relaunch chatter, admin-scope creep, Evergreen exception handling, or founder-default behavior.

Jake is starting the Compass prototype today off the internal owner tool. I want the execution criteria written down before this drifts into dashboard or customer-proof territory.

Jake DM: I can start the Compass prototype today off the internal owner tool. Proposed fields: - prompt_type: renewal-risk/admin-friction or post-first-live-sync - account - owner - next_action - evidence_source - caveat - owner_action_status Questions: 1. Should rejected rows be hidden, or do we keep them visible somewhere? 2. Does owner_action_status need more than not_started / done / blocked? 3. Do we need a due date field for September, or will that create fake precision? I am not building a dashboard or customer-facing surface.

Turn this into a short internal execution note and send Jake, Anna, Sarah, and Leo the acceptance criteria: rejected-thin rows stay visible, no fake due-date precision, no dashboard, no customer-facing proof, and owner_action_status has to stay tied to evidence and caveats.

Sarah sent rough account-thread examples for Compass. One Evergreen row looks usable, one Acme row is tempting but wrong, and one Evergreen admin-friction row is too thin. I want this cleaned before people start treating narrative as evidence.

Sarah DM: Rough Compass examples for prototype input: 1) Evergreen Bank / post-first-live-sync Source: Aug 14 account note from me after admin office hours. Quote: "Once first live sync completed, treasury ops asked who should own the next source because the invited teammate was unsure whether they were expected to reconnect or wait for an admin." Owner: me. Follow-up: Aug 15 I sent the safe setup-guide line after Priya's copy fix. Caveat: this is still support/account-thread follow-up, not customer behavior proof. 2) Acme / renewal-risk-admin-friction maybe? Source: July export timeout thread. Owner: support + Leo first read. Follow-up happened, but it was API v2 export fix and customer rerun, not a Compass prompt. Caveat: probably should not count; including because it looks like an owner action if someone squints. 3) Evergreen Bank / admin-friction Source: I have a note saying "another department may ask about admin policy" but I don't have the original customer date in the row yet. Owner would be me/Devon depending on commercial framing. Caveat: too thin until I attach the actual customer source.

Make a cleaned internal note and reply to Sarah: keep the Evergreen post-first-live-sync row as usable with caveat, reject the Acme export row as Atlas context, and mark the undated Evergreen admin-policy row as too thin until the actual source/date is attached.

Anna sent the first pass on the Compass evidence threshold. I want this locked before implementation bakes in the wrong interpretation.

Anna DM: Compass evidence threshold draft: For September prototype readout, I want to count: - completed owner action only if there is a named owner, dated next action, evidence source, caveat, and completion status - rejected row if the account note lacks a source/date or the prompt sighting is too vague to act on - waiting row if the owner action is real but depends on live customer follow-up Do not count: - raw prompt sighting - cleaner copy or better prep script by itself - Acme export timeout or other Atlas support work - customer-facing behavior change unless we actually see it, which I do not expect from this prototype Question: can Jake make rejected rows first-class in the prototype instead of hiding them in a notes field?

Turn that into a short internal reference note and DM Anna and Jake that rejected rows need to be first-class status. The readout stays internal owner-action evidence only.

Kara finally sent a safe Compass paragraph. The relaunch and customer-proof stuff is gone, and I want to stop the iteration loop here.

Subject: Revised Compass paragraph — proof language removed Morgan, We removed the relaunch / customer proof wording. Current version: "Scaffold is using Q3 operating work to tighten internal account-owner follow-through around renewal-risk/admin-friction signals and post-first-live-sync expansion prompts. The work is being evaluated as internal owner-action evidence, with source notes and caveats preserved. We are not positioning this as a customer-facing launch, market proof, or a staffing trigger." We also removed the draft line that said "Compass is proving repeatable customer growth." Let me know if this is finally safe to use in the internal planning memo. Kara

Reply in-thread with a concise approval. Condition: Kestrel does not reintroduce customer-proof, relaunch, GTM, staffing, or fundraising language around Compass.

Sofia wants to know if anything in the rest of August changes the Q3 operating read materially. I want to keep her current without turning this into a board pre-read or Series C warm-up.

Subject: Q3 operating evidence — August read? Morgan, Do you expect anything in the rest of August that changes the Q3 operating read materially? I am not asking for a board pre-read yet; I just want to avoid surprises if Compass, Evergreen, or Atlas/Mercury reliability is moving faster or slower than expected. Sofia

Reply short: Acme is closed as Atlas reliability/routing evidence, Compass is internal owner-action prototype work, Mercury copy quality improved without admin-scope change, Evergreen is still a strong but account-specific signal unless we get a real repeatability read, and none of this is a Series C trigger.

Sarah forwarded a new Evergreen ask. Another internal department wants to try the current Mercury flow under the existing annual Growth account, and Evergreen is asking again for a custom admin-policy date. I want those separated immediately so nobody accidentally promises roadmap.

Forwarded Evergreen email: Subject: Adding treasury operations team to existing Growth account Sarah, We have a second internal group — treasury operations — that would like to try the same Mercury setup path under our existing annual Growth account. They want to create a workspace, invite the teammate who will own the first source connection, connect the first source, and confirm first live sync. Can this be added under the current annual account, or does this need to be treated as a new department / expansion first? Separately, can you include the expected date for the custom admin policy controls we discussed earlier? Our internal rollout tracker is asking when we can lock invite/source-owner permissions by department policy rather than relying on the current basic roles. Thanks. Evergreen Bank platform operations Sarah note to me: I have not answered. I can say current setup guide/source-owner path is clearer after Priya's copy change, but I do not want to imply an admin-policy date or expansion. Should I route Devon on the commercial part and Jake on current Mercury scope?

Create a quick internal intake note and DM Sarah, Jake, and Devon: the added department can only be evaluated as a current-flow test if Devon confirms commercial framing; custom admin policy/date is not promised; Jake should confirm current shipped commitments are workspace setup, teammate invite, source connection, and first live sync.

Jake and Devon both replied. Jake bounded the product scope, Devon is fine with commercial framing under the existing annual Growth account. I want Sarah to get exact wording now so this doesn't sprawl.

Jake: Current Mercury scope for this is only: - workspace setup - teammate invite / source-owner path - connect first source - first live sync No custom admin policy. No granular role editor. No department-level permission lock. No SSO change. No Atlas/auth scope. Priya's copy change only made the current source-owner action clearer. Devon: Commercially I am okay with treasury ops testing the current flow under Evergreen's existing annual Growth account. I would not call it a paid expansion or new department package. I also do not have an advanced-admin add-on/date to quote. Fine for me to be referenced as commercial framing only; I should not become product owner for the thread.

DM Sarah a customer-safe reply she can send: treasury ops may test the current shipped flow under the existing annual Growth account; the test is limited to setup, invite/source-owner path, first source connection, and first live sync; we are not providing a custom admin-policy date or committing advanced admin controls.

Evergreen accepted the bounded second-department test under the existing annual Growth account, and Sarah held the line again on admin-policy dates. I want the record updated now so this doesn't get rewritten later as expansion or advanced-admin scope.

Sarah update: Evergreen update: I sent the bounded line yesterday. They are good with treasury ops testing the current Mercury path under the existing annual Growth account. Their summary back: - They will use current workspace setup. - They will invite the teammate who owns first source connection. - They will connect one source and confirm first live sync. - No SSO change requested for this test. They did ask once more whether I could put a custom admin-policy date in their procurement tracker. I said we are not providing a date or commitment for custom admin policy / granular role editor / broader advanced-admin controls, and that this test is only for the current shipped path. Devon stayed in commercial framing only. Jake's scope line was enough.

Create the closeout record, update Evergreen's CRM row and the Q3 operating-evidence slide, and send owner notes to Sarah, Jake, and Devon. Durable fact needs to be exact: second department may test current flow under the annual Growth account; no advanced-admin promise; Devon stays commercial framing only.

Evergreen is closed out and I'm leaving now. Text Jamie that I should be home around 6:25 and I'm keeping tonight offline unless something is actually on fire.

Jake, Leo, and Anna all checked in on the Compass prototype shape this morning. Field and status alignment looks right, Leo doesn't see a new data-store requirement, and I want the progress captured without anybody treating it like a result.

Jake: I have the prototype branch wired in the internal owner tool with these fields: account, prompt_type, owner, next_action, evidence_source, caveat, owner_action_status. Status values are not_started, done, rejected_thin_source_or_date, blocked_waiting_customer_follow_up. No due date field. No dashboard. Leo: Data seam check: this can use existing account-note references + a lightweight status table in the internal owner tool. I do not see a new platform/data-store requirement. Need a source freshness warning if the account note is older than the owner action. Anna: This matches the evidence threshold. Please do not let done count without source + caveat. Rejected-thin needs to show up in the readout count, not disappear as invalid data.

Make a short implementation-status note and DM Jake: good progress, add Leo's source-freshness warning, keep rejected-thin visible, and do not call this a readout result yet.

Small Evergreen update since Tuesday: they accepted a second-department test of the current Mercury flow under the existing annual Growth account. Useful signal, but still inside one customer and with no admin-scope expansion.