Morgan Chen

Can you place the usual Friday Lemongrass order for me + Jamie and send me the ETA / confirmation when it comes through?

Just noticed Kenji's Apr 30 BCI follow-up. Send him a warm personal reply thanking him for the paper and saying coffee would be great -- Tuesday late afternoon or Thursday morning next week both work. No company pitch, and don't put this in the CRM.

Can you check my calendar for this morning? If 10am-12pm PT is open, block it as protected time for Mercury operating work -- Jake's weekly agenda + blockers. Don't move anything if there's already a conflict.

Jake's asking which observability option we rejected so Lisa doesn't reopen it. Send him a concise internal reply: Honeycomb is the choice, Datadog is the one we passed on, and include the actual rationale from the decision note. Keep it short.

Rishi's old standup question is stale at this point. Can you check the open Atlas PRs and, if the verifier-extraction PR is still open, comment there asking him to call out any remaining JWT cutover risk on the PR? Don't revive the missed standup thread.

Greg's contractor-scope email is still sitting there. Send him a brief external reply asking for the May deliverable list and billing assumptions in summary form only. Keep roadmap detail out of it and use the normal short external sig.

Need a thinking pass, not a send. Jake asked a confidential HR/comp question and I want to answer cleanly in 1:1. Help me frame it so I acknowledge the question, point him to the framework/process, and don't accidentally imply any comp promise.

Before Jake's Mercury weekly, check open Mercury PRs. If the feature-flag rollout PR is still open, post a comment asking for the rollout checklist and the customer/admin-blocker implications before prod. Short, not a lecture.

Use these raw notes from Mercury weekly. Mercury weekly — raw notes Tue May 9, 2023 Attendees: Morgan, Jake, Anna, Priya, Devon - Keep this meeting on product/operating work only. No investor / deck detour. - Jake wants one visible weekly list: blocker, owner, next decision. - Current launch blockers are mostly clustered, not scattered: - admin invites fail in some org setups; repro seems tied to inviter role + pending seat/account state - activation instrumentation still has a gap between workspace created and first real key action - account-level permissions are confusing enough that admins hit a wall before setup completes - Priya: activation/onboarding UX can probably remove one whole dead-end, but she needs engineering call on what is real product friction vs permissions bug. - Anna: weekly retention cut is useful only if freshness is explicit; she will own keeping the cut current and flagging what changed vs backfill/noise. - Jake: owns admin sequencing. Need one answer on what order we force for invite, permissions, first admin task. - Morgan: keep Mercury weekly as the operating spine instead of rewriting this into a story every Friday. - Devon: okay with that, but wants the retention work connected to actual launch blockers rather than living in a side doc. Owner calls - Anna -> retention freshness / instrumentation sanity / changed-cut notes - Priya -> activation UX pass, especially first-run path + invite/setup dead ends - Jake -> admin sequencing, permissions dependencies, what can be cut vs launch-blocking Questions to take forward - Which admin steps are true launch blockers vs annoying but deferrable? - Do we gate rollout on fixed account-level permissions or can we ship a narrower path? - What event do we trust as the activated point this week? Open blocker list leaving the meeting 1. admin invites 2. activation instrumentation 3. account-level permissions Loose next steps - Jake to bring proposed admin sequence. - Priya to show where UX can remove confusion without needing backend changes. - Anna to mark fresh vs noisy numbers on the retention cut. - Morgan to keep Friday wrap out of this and keep this out of investor narrative. Create/update the saved note as `Mercury weekly — May 9 operating list`, then post a short internal owner list: Anna on retention freshness, Priya on activation UX, Jake on admin sequencing. Include the blocker list too -- admin invites, activation instrumentation, account-level permissions.

Anna sent the retention cut. From: Anna Martinez To: Morgan Chen, Devon Hayes Date: Tue, May 9, 2023 1:18 PM Subject: May 9 retention cut export Pasting the export below since the sheet permissions were being weird. [signup_month_cohorts] cohort,accounts_started,wk1_activation_rate,wk4_active_rate,m2_logo_retention,m3_logo_retention,expansion_accounts,notes 2022-11,28,71%,61%,57%,54%,6,stable sample 2022-12,31,68%,58%,55%,53%,5,holiday timing but sample is okay 2023-01,37,73%,63%,59%,56%,7,best post-onboarding-fixes cohort so far 2023-02,29,69%,60%,57%,55%,4,stable enough; some expansion still lagging into later periods 2023-03,18,72%,56%,50%,n/a,1,still maturing 2023-04,21,67%,52%,n/a,n/a,0,too early for a retention read [weekly_activation_counts] week_start,signups,workspace_created,invite_sent,first_admin_action,first_key_action_within_7d,activation_rate_7d,notes 2023-04-03,12,11,8,7,7,58%,one enterprise start landed midweek 2023-04-10,9,8,6,4,4,44%,small N 2023-04-17,14,12,9,8,8,57%,cleaner week 2023-04-24,11,10,6,5,5,45%,invite drop-off is obvious here 2023-05-01,8,7,5,4,3,38%,small N and likely to move a bit with late event arrival [sample_size_notes] - Weekly slices swing hard when signups are under about 12. - One team-plan customer can move invite_sent and first_admin_action by itself. - April cohorts are not mature enough for external retention claims. - Monthly cohort shape is good enough for operating direction; weekly activation is better for spotting where onboarding breaks than for presentation. - I left out the current week starting 2023-05-08 because it is too partial. Don't create a doc or send anything. I just need the read: what is safe to use for internal operating decisions this week, and what is too noisy for board/external charts?

AP digest is still showing the Foundry April invoice pending, but I think there's already a paid copy. Please inspect the invoice records and flag the pending one for manual review as a possible duplicate. Do not pay it.

Mom texted. SMS thread Tue, May 9, 2023 Mom: How is the new apartment? Are you sleeping enough? How is Kibo? Draft me a short reply I can send myself -- apartment/Kibo/Jamie tone, reassure her work is busy but fine. Don't mention late nights or stress.

May 12 offsite is coming up fast. Create a working agenda doc titled `Mercury product offsite — May 12 working agenda`. Keep it built around decisions/cuts/owners/blockers: launch blockers kickoff, activation/onboarding cuts, admin blocker triage, retention instrumentation with Anna, owner assignments, and inputs for the next Mercury weekly.

Customer Success sent the blocker digest. From: Customer Success Date: Wed, May 10, 2023 8:21 AM PT To: Morgan Chen, Jake, Priya, Anna Martinez, Devon Hayes Subject: Mercury blocker digest — May 8-9 (account names removed) Account names removed as requested. This is just the blocker pattern digest from the last two days, not a proposed plan. Top themes across the May 8-9 calls/tickets - Activation dead ends: 7 accounts - Admin invite failures / confusing invite states: 5 accounts - Onboarding confusion before first value: 6 accounts - Several accounts showed up in more than one bucket, so counts are not additive. 1) Activation dead ends What we heard - New admins are getting through workspace creation and then stalling on the first "connect / import / continue" decision. - When the connector step fails or is skipped, people are landing in an empty state that reads like setup is still in progress but does not give a clear next action. - In a few cases the champion assumed the product was not fully enabled yet and stopped instead of trying another path. Observed patterns - The setup checklist implies a single correct order, but users are hitting paths where that order breaks. - "Connect data" is reading as mandatory even for teams that expected to start with sample data or manual setup. - Error text after a failed connection attempt is too generic; users are not sure whether to retry, switch methods, or ask an admin. Representative notes - "I connected, got bounced back, and now I can't tell if it saved anything." - "We wanted to invite the team first and come back to the source later, but the flow kept pushing us back into setup." - "The page looked unfinished, so we waited for your team instead of clicking around." Impact - 3 accounts did not complete first-session setup without manual CSM intervention. - 2 accounts asked whether the rollout was still in beta because the post-create state looked incomplete. 2) Admin invite failures / confusing invite states What we heard - Admins are sending teammate invites and not trusting the result. - Some invites appear to send but recipients never receive them. - Some admins see a "pending" state that stays around even when the teammate already exists or has already tried to join. Observed patterns - Duplicate-invite handling is not obvious from the UI. - Domain / role restrictions are surfacing late, after the admin thinks the invite already went out. - The distinction between workspace admin vs other roles is unclear at the exact moment people are inviting their first teammates. Representative notes - "It said pending, but the person was already in there on their side." - "I invited two people and got no confirmation other than the spinner stopping." - "If this is blocked because of permissions, say that before I type the emails." Impact - 2 accounts paused rollout until their internal admin could verify whether invites had actually been sent. - 1 account attempted the same invite flow three times and created support noise because nobody trusted the state. 3) Onboarding confusion before first value What we heard - Teams are not getting to a crisp first success moment fast enough. - Users are reading the setup screens as product documentation rather than guided onboarding. - The copy around roles, first project/setup object, and team invite timing is still muddy. Observed patterns - People do not understand whether they should invite teammates before or after the first source/project is set up. - "Admin" terminology is overloaded; some people are reading billing/admin/account admin as the same thing. - The checklist is doing too much explaining and not enough steering. Representative notes - "I wasn't sure what counted as done, so I kept clicking around." - "We invited the team before anything was ready and then everybody landed in a half-empty workspace." - "I needed one clear next step, not three almost-right ones." Impact - 4 accounts reached support with some version of "not blocked exactly, but not confident enough to proceed." - In 2 cases the team got to value only after a live walkthrough. Cross-cutting note - These are not mostly feature requests. The common thread is uncertainty during setup: users do not know which path is valid, whether an action succeeded, or who should do the next step. Raw asks that came up more than once - Clear fallback path if connector step fails - More explicit confirmation after invite send - Plain-language role copy at invite time - Checklist that adapts instead of trapping users in one order - Better "what do I do next" state after workspace creation If useful, Customer Success can break this back out by funnel step next, but the main signal from May 8-9 is that setup is failing more from ambiguity than from outright missing capability. Update the Mercury operating/offsite materials so these show up as decisions we need to make, not color in the story: activation dead ends, admin invite failures, onboarding confusion before first value.

Priya exported the Figma comments. Figma comment export File: Mercury / activation + onboarding Exported by: Priya Exported at: Wed, May 10, 2023 11:06 AM PT Included: component notes + unresolved comment threads Page: Activation / onboarding v4 Frame: 01 — Workspace created / first-run checklist Component notes - Goal of this screen is to give one obvious next move, not present every setup option. - Checklist should collapse to the shortest path for a new admin. - If a user skips connect/import, we need an intentional fallback state rather than a blank or half-complete dashboard. - Role copy here is still placeholder and should not ship as-is. Unresolved thread C-118 Priya — May 9, 2:14 PM The dead-end here is still real. If connection fails or is deferred, we dump people into a state that looks like setup is broken instead of incomplete. Need product decision: do we route to sample/manual path, or keep them on checklist with a clear alternate CTA? Jake — May 9, 4:02 PM We can support an alternate CTA, but only if we stop assuming the connector path always produces org-ready state. Right now backend state is cleaner when connect happens first. Priya — May 9, 4:18 PM That's exactly the point of the comment. If the backend assumption forces this order, call that out as implementation constraint, not UX preference. Status: unresolved Unresolved thread C-121 Morgan Chen — May 9, 5:07 PM The copy still sounds like there is one correct setup order. We keep hearing from customers that they want to invite a teammate first or at least understand roles first. Priya — May 9, 5:14 PM Agree. I can rewrite the checklist copy, but if the flow actually breaks when they invite before connect, that is not a copy fix. Status: unresolved Frame: 02 — Connect data / import choice Component notes - "Connect data" and "Start with sample data" should feel like two legitimate paths, not one primary and one apology. - Error treatment should be inline and next-step oriented. - If auth fails, user should not lose all context from the prior step. Unresolved thread C-127 Priya — May 9, 2:33 PM Current error block is too vague. "Something went wrong" is not enough when the user needs to know whether to retry, switch methods, or ask their admin. Jake — May 9, 4:11 PM I can surface the actual failure class for permission-denied vs provider-timeout. I do not want to expose raw provider errors. Priya — May 9, 4:16 PM Not asking for raw provider errors. Just need actionable states. Permission issue, temporary retry issue, or choose another path. Status: unresolved Unresolved thread C-130 Priya — May 9, 6:02 PM If sample/manual is a real option, the button placement needs to stop looking secondary/unsafe. Right now it reads like "continue without doing the real thing." Status: unresolved Frame: 03 — Invite teammates Component notes - Confirmation state must be explicit. - Duplicate invite handling needs visible system behavior. - Role labels should use plain language at the moment of action. Unresolved thread C-141 Priya — May 9, 2:48 PM This spinner/return-to-same-state pattern is why admins think the invite did not send. We need either a sent state with email list preserved or a row-level confirmation state. Jake — May 9, 4:24 PM There is a backend 409 on duplicate or existing-member cases. UI is swallowing it and collapsing back to pending. Priya — May 9, 4:29 PM That's useful. Then the design question is whether "already in workspace" and "invite pending" should be different visible outcomes. I think yes. Status: unresolved Unresolved thread C-146 Morgan Chen — May 9, 5:22 PM "Admin" here is still doing too much work. People are reading this as billing authority / account owner / workspace setup owner all at once. Priya — May 9, 5:31 PM I can split the role copy, but I need the actual permission boundary from engineering so we do not fake clarity. Status: unresolved Frame: 04 — Permissions / restricted action state Component notes - This state should explain why the user is blocked and who can unblock them. - Avoid generic lock-icon language. - We need a plain-language fallback when the user is the wrong person to finish setup. Unresolved thread C-155 Priya — May 9, 3:05 PM This screen is visually clean but still too abstract. "You don't have access" is not enough. It should say what they can still do and who needs to step in. Jake — May 9, 4:37 PM Fine from my side as long as we do not imply permissions that do not exist yet. Status: unresolved Frame: 05 — Post-setup empty state / first value Component notes - First value state should prove the workspace is real and usable. - Empty state should not read like a broken dashboard. - We may need a more opinionated first recommended action. Unresolved thread C-163 Priya — May 9, 3:22 PM This is prettier than before but still not a satisfying first-success moment. If users land here after partial setup, it looks like the app forgot what they just did. Morgan Chen — May 9, 5:41 PM This matches the customer-success feedback almost exactly. Priya — May 9, 5:44 PM Then I want the offsite to separate two things: are we choosing this state on purpose, or is this what the current implementation leaves us with? Status: unresolved General file note from Priya - The file is not blocked on visual cleanup. - The unresolved issues are mostly about where product intent ends and implementation constraint begins. - Biggest open questions from my side: valid setup order, explicit invite outcomes, and whether fallback paths are real product paths or edge-case patches. Give me five sharp offsite questions that separate actual design choices from implementation blockers. Don't send anything.

Can you look at the May 19 customer-visit calendar entry? It says me, Devon, and Jake are traveling SFO-JFK, but before anything gets booked I want a short checklist of what we still need: departure window, return window, who is actually traveling, and hotel constraints.

Devon marked up the offsite agenda. Document comments export Document: Mercury product offsite — May 12 working agenda Commented by: Devon Hayes Date: Thu, May 11, 2023 8:07 AM — comment on title Suggested title feels too tactical. Could this be framed more like "Mercury launch working session" so people walk in with the bigger picture in mind? 8:09 AM — comment on opening section: "Launch blockers and cuts" I think we need 15-20 min up front on the story of why Mercury matters now, otherwise the rest can feel like a random blocker sort. Even internally, the team needs the spine. Suggested insert: Section 0 — Why Mercury / why now - What changed in the last 60 days - Why activation + admin pain is the right wedge - What we need true by the end of Q2 8:12 AM — comment on section: "Activation / onboarding cuts" Can we anchor this in retention effect, not just UX cleanup? I would add one short slide or chart here showing pre-fix vs post-fix shape after onboarding changes, even if it is still directional. Suggested slide: Slide 1 — Retention shape after onboarding fixes Subtitle: Not "solved," but reading as product effect instead of noise 8:14 AM — comment on section: "Admin blocker triage" This should probably connect to expansion/admin pain, not sit as a loose bug bucket. If we do not say why admin issues matter, we will over-index on one-off fixes. Suggested slide: Slide 2 — Admin friction as expansion blocker - Invite reliability - Permission confusion - Account-level controls - Why these are gating broader rollout inside teams 8:17 AM — comment on section: "Retention instrumentation with Anna" I would avoid making this sound like a data-cleanup side quest. This is part of the launch narrative. Maybe present it as: what is signal, what is caveat, what needs to be true before we tell the story more broadly. Suggested insert: Section — Retention readout and narrative guardrails - Which cuts are solid enough to operate on - Which cuts are still moving because of mapping/instrumentation - What language we can use confidently by June 8:21 AM — comment on section: "Owner assignments" Would be useful to end with a single page that says: - what we are launching - what we are cutting - who owns each blocker - what evidence will tell us the launch is working 8:25 AM — comment on end of doc I know we do not want to turn this into a board deck, but I also do not want us to miss the chance to get everyone aligned on the actual launch story. A few lightweight slides would help. Draft narrative language from Devon - Mercury is the point where our retention curve starts to read as product effect rather than interpretation. - The through-line is simple: activation gets people to first value faster, admin fixes keep rollout from stalling, and cleaner instrumentation tells us whether the product changes are actually working. - If we leave the room with only a blocker list, we will have done triage, not alignment. - By the end of Q2, we should be able to say the launch removed the biggest self-inflicted friction in setup and admin adoption, even if the full retention picture is still maturing. Proposed lightweight slide list 1. Why Mercury, now 2. Retention shape after onboarding changes 3. Activation + admin pain as one launch problem 4. What is shipping vs what is being cut 5. Owners / evidence / next checkpoint Inline note on timing If we keep the current agenda order, I would still put 10 minutes at the top for context and 10 minutes near the end for the launch readout. Otherwise Anna's section will feel disconnected from the rest. Update the saved agenda doc, but keep the line: this is an operating offsite, not a launch narrative or deck rehearsal. Strip the narrative/deck sections and preserve only decision points, cuts, owners, and blockers.

Agenda is settled enough. Check May 12 and if there's no offsite hold already, create an internal calendar event 10am-4pm PT titled exactly `Mercury product offsite — decisions, owners, launch blockers`. Attendees: me, Devon, Jake, Priya, Anna. Keep the agenda in the saved doc, not pasted into the calendar body.

Anna just sent the mapping-change note. From: Anna Martinez Date: Thu, May 11, 2023 1:36 PM PT To: Morgan Chen, Devon Hayes Subject: weekly retention cut moved after two activation mapping fixes Quick note because the weekly rows changed more than I expected after this morning's cleanup. What changed 1) "activation_complete" was only picking up successful connector auth on the new path. - It was missing accounts that reached a usable workspace through sample data / manual-import setup. - Those accounts were showing up as created but not activated. 2) "first_team_action" was mapped to invite-email enqueue rather than a persisted teammate-added event. - That means some failed or duplicate invite attempts were being treated as real downstream team activation. - I corrected this to the persisted teammate-added / member-visible state. Why the weekly cut moved - Fix #1 increased the count of accounts that should qualify as activated in the last two weekly cohorts. - Fix #2 removed a small number of false positives from the team-activation side. - Net result: the denominators and early follow-on rows shifted. Updated rows | signup week | prior activated accts | corrected activated accts | prior wk1 retained | corrected wk1 retained | prior wk2 retained | corrected wk2 retained | |-------------|-----------------------|---------------------------|--------------------|------------------------|--------------------|------------------------| | 2023-04-17 | 18 | 22 | 61% | 55% | 44% | 47% | | 2023-04-24 | 14 | 19 | 57% | 53% | n/a | n/a | | 2023-05-01 | 9 | 11 | too early | too early | n/a | n/a | Interpretation - This does not change the broader quarterly read materially. - It does change the weekly story enough that I would not freeze any weekly chart into something board-facing or external. - Internally, the weekly cut is still useful for operating decisions if we treat it as directional and annotate mapping changes. My current view - This is primarily an instrumentation correction, not evidence that the product suddenly got better or worse overnight. - It does touch launch-readiness in one way: if we are using weekly activation/retention rows to decide cuts at the offsite, we should be explicit about which rows just moved because of event hygiene. - I do not think this by itself is a reason to block launch work, but I also would not let anyone cite the old weekly numbers as settled. Sanity check I ran - Monthly / quarterly cohort shapes barely move. - The movement is concentrated in the newest weekly cohorts where sample size is already small. - That is exactly where the mapping error would have the biggest visible effect. Recommendation for discussion - Use the corrected weekly rows for internal operating conversation. - Mark them as updated after event-mapping fixes. - Keep quarterly cohorts as the only thing we would defend outside the room. If helpful I can bring a one-pager tomorrow with "signal vs caveat" annotated directly on the tables. Help me reason through how to treat this tomorrow: launch blocker, instrumentation task, or reporting caveat? No doc, no reply -- I need the judgment.

Post a short prep note in Discord for the Mercury offsite. Point people to the saved agenda and tell them to come with: cuts they will actually make, blockers they need decided, and owner proposals. No investor framing.

Greg replied with proposed May scope. From: Greg Shipman Date: Thu, May 11, 2023 9:18 AM PT To: Morgan Chen Cc: Sarah Kim <sarah@atlas-test.com> Subject: Re: May deliverables + billing assumptions Morgan — understood on keeping this high-level. Summary below. Proposed May deliverables 1. Admin invite reliability pass - clean up duplicate-invite handling - address the no-confirmation / stuck-pending behavior - close the most visible failure states in the current invite flow 2. Setup checklist cleanup - tighten the first-run checklist so it is less easy to stall in setup - simplify a few pieces of onboarding copy where users are misreading next steps - basic polish pass on the post-create state so it feels less half-finished 3. Remaining dashboard/render defects already in the queue - close out the current small batch that we have already discussed - no net-new module work implied here 4. Optional follow-on if you want us to carry it this month - early groundwork on permissions/admin-console cleanup - light pass on adjacent activation friction if we have room after the items above Billing assumptions - baseline assumption is 72 engineering hours + 16 QA hours + 4 PM/review hours for items 1-3 - assumes one consolidated review cycle from your side, not multiple staggered rounds - assumes no major scope change once work starts - assumes Scaffold provides test access / repro details promptly where needed - if you want item 4 included, I would treat that as incremental and likely another 30-40 hours depending on how far you want to take it A couple of framing notes from my side - Items 1 and 2 are linked in practice, so I would expect some spillover between them at the implementation level even if they are tracked separately for billing. - The permissions/admin-console piece is less a fixed deliverable than a sensible next area if we want to prevent the same setup/admin friction from resurfacing. - If you would rather keep May extremely tight, we can confine it to items 1-3 and leave the follow-on discussion for later. Happy to tighten wording further if you want this broken into "in scope" vs "possible next" more explicitly. Greg -----Original Message----- From: Morgan Chen Sent: Mon, May 8, 2023 4:42 PM PT To: Greg Shipman Cc: Sarah Kim <sarah@atlas-test.com> Subject: May deliverables + billing assumptions Greg — can you send the May deliverable list and billing assumptions in summary form only? Please separate what is clearly in scope from any assumptions or guesses about follow-on work. I do not want roadmap detail in email. Morgan Chen · Scaffold Draft a concise response for my review only. Accept only the clearly scoped items, ask him to separate billing assumptions from roadmap guesses, and keep it summary-only. Don't send -- this one is touchy.

Final logistics thread for the Mercury offsite is here: Subject: Fwd: Mercury offsite - final room + lunch confirmations From: Morgan Chen Sent: Thu, May 11, 2023 5:02 PM To: Morgan Chen Forwarding the final logistics thread so I have it in one place for tomorrow. ---------- Forwarded message --------- From: Sarah Kim Sent: Thu, May 11, 2023 4:41 PM Subject: Re: Mercury offsite - final room + lunch confirmations To: Morgan Chen, Devon Hayes, Jake, Priya, Anna Martinez All set for tomorrow. Key logistics pulled to the top so nobody has to hunt: - Room is confirmed 8:30am-5:30pm for 5 people. - Property ops confirmed business-grade internet directly, not just website marketing copy. - Tava Kitchen delivery is confirmed for 12:10-12:25pm. - Whiteboards, HDMI, and a hardline are available in the room. Forwarded confirmations are below. Sarah Begin forwarded message: From: Lena Ortiz, Property Operations Date: Thu, May 11, 2023 2:16 PM Subject: Re: Internet confirmation for May 12 booking To: Sarah Kim, Morgan Chen Hi Sarah and Morgan, Confirming directly from property operations for the May 12 booking: Room - Bay Room reserved Friday, May 12, 8:30am-5:30pm - Setup: conference layout for 5 - Two rolling whiteboards in room - One wall-mounted display with HDMI cable - Extension cord and power strip available on request; I have already asked the floor team to place them in the room before arrival Internet - The room is served by the building's dedicated business fiber circuit - Standard working speed on that floor is 300/300 Mbps symmetric - Hardline ethernet port is live in the room - Separate guest Wi-Fi SSID is available with no captive-portal splash page - This is the same circuit our resident companies use during the workday; it is not shared event Wi-Fi - We have had no reported outage on that circuit in the last 60 days - Building ops will be onsite starting at 8:00am if anything needs to be reset If your team is doing a working session with live product demos, this setup is appropriate for that use. Best, Lena Property Operations On Thu, May 11, 2023 at 11:08 AM Sarah Kim wrote: Hi Lena — one specific ask before we finalize: can you confirm the room has business-grade internet from the property ops side? We need a real confirmation from ops, not the website listing. Team will be doing product work and demos. Thanks, Sarah End forwarded message Begin forwarded message: From: Tava Kitchen Catering Date: Thu, May 11, 2023 3:07 PM Subject: Re: May 12 lunch delivery for Scaffold team To: Sarah Kim Cc: Morgan Chen Confirmed for Friday, May 12. Delivery window - Arrival window: 12:10pm-12:25pm - Drop-off: front desk, labeled for Scaffold / Bay Room - Disposable serving set included Headcount - 5 meals total Menu 1. Harissa chicken bowl 2. Harissa chicken bowl 3. Falafel tahini bowl 4. Green goddess chicken wrap 5. Avocado + hummus wrap Sides / extras - Mixed greens salad for the table - Kettle chips - Sparkling water assortment - Cookies (separate box) Notes on packaging - Each item will be individually labeled - Vegetarian items marked clearly - Dressing and sauces packed on side If the team needs the delivery a little earlier because the session starts before lunch, reply by 6:00pm and we can try to move it up slightly. Otherwise we will keep the 12:10-12:25 window. Thanks, Tava Catering Team On Wed, May 10, 2023 at 4:52 PM Sarah Kim wrote: Hi team — placing the lunch now for Friday's Mercury working session. Please confirm: - 5 people - Delivery window around 12:15 - Mixed menu is fine - Need vegetarian items labeled Thanks, Sarah End forwarded message On Wed, May 10, 2023 at 5:18 PM Morgan Chen wrote: Thanks. Keeping this one day and product-deep, so the only thing I care about on logistics is that the room actually works and lunch shows up on time. Morgan Please save a one-day run-of-show note from this. Keep the room product-deep: confirmed internet and lunch logistics, short demos before working blocks, and no investor-facing narrative in the room.

Devon is already nudging the kickoff toward a cleaner investor story. Give me a short spoken redirect I can use after the break: Sarah at Founders Fund was right that this should stay one day, but the reason is focus, not performance. This is working time for the product team, not a rehearsal for investors.

The first demo block helped, but open Q&A is starting to go dead-air. Give me quick facilitation prompts to turn the afternoon into three working blocks: activation, expansion, admin pain. I want decisions and owner calls, not wandering opinions.

Here’s the closeout whiteboard dump: Mercury offsite closeout whiteboard notes Photo transcription, 5/12 late afternoon [center top] MERCURY OFFSITE CLOSE not a deck ship work / owners / cuts / blockers weekly spine = Jake if discussion turns into narrative polish -> cut it [top right] Launch bar: - activation without founder handholding - expansion path visible in product, not in slides - top admin pain reduced enough that calls stop repeating the same 3 complaints - one blocker owner each [left board] ACTIVATION Owner(s) - Priya -> path + copy + screens - Jake -> implementation / flag / rollout sequence - Anna -> event definitions + success cut Need by next review - single primary CTA = connect first source - invite teammates becomes optional after first value, not before - sample data only if no real event in first session - define activation event chain in plain English + actual events Blockers - billing gate timing still unresolved * before first real run = probably too much friction * after first value = better, but need abuse cap / usage limit - event names still fuzzy; Anna needs exact handoff, Jake won't wire to vibes - copy on role setup still abstract; admins don't know what changes later vs now - feature-flag rollout criteria not written anywhere Scope cuts - no template gallery for v1 - no SSO in first-run path - no advanced permissions setup during activation - no logo/upload/custom branding in first-run Notes in margin - "first value" means connected source + first successful action, not invited teammate - stop debating empty-state illustration - do not reopen entire settings IA here [middle board] EXPANSION Owner(s) - Jake -> overall - Priya -> upgrade moments / UI - Anna -> usage threshold sanity check / measurement What stays in - expose upgrade trigger where usage actually hits limit - one clean in-app path from cap hit -> plan/owner action - admin sees why workspace is constrained - light copy only, no packaging essay Blockers - Mercury feature-flag rollout still ambiguous * who flips it * merge criteria * rollback note * target date - usage threshold cut is noisy; need one number the team will actually use - upgrade copy still split between product/admin language and billing language Scope cuts - no annual-plan flow work for launch - no seat-management overhaul - no contract-specific enterprise branching in product - no pricing-story work in this room Notes in margin - expansion path has to exist even if monetization details evolve later - if it requires sales explanation every time, it is not launch-ready [right board] ADMIN PAIN Owner(s) - Jake -> backend / account actions / audit surfaces - Priya -> admin home + settings grouping - Anna -> tag incoming issues / severity / frequency Repeated pain to actually fix 1. permissions/roles unclear 2. bulk user actions too manual 3. no clean audit trail for "who changed what" 4. wrong person gets admin/billing noise 5. settings scattered across too many places Blockers - audit log scope unclear: what events are mandatory for launch? - CSV/import failures still bad at telling admins what broke - permission model explanation still too hand-wavey for security-conscious customers - notification routing rules not explicit enough Scope cuts - no full settings redesign - no granular notification matrix - no admin analytics dashboard - no custom role builder Bottom line for admin block - fix the repeated pain, not every admin complaint anyone has ever had - if a problem only matters after launch scale, write it down and move on [bottom strip across boards] Cross-cutting - Jake runs weekly from this, not Morgan - Morgan only pulled in for actual scope calls / unblock - Devon can help pressure-test logic, but not turn it into investor copy - every owner comes to weekly with: what moved / what slipped / what we cut - customer examples > opinions - launch blockers list stays short and named [boxed in red] OPEN BLOCKERS TO CARRY FORWARD - billing gate timing - activation instrumentation definitions - Mercury feature-flag rollout owner + merge/rollback criteria - admin audit-log minimum scope [small note in lower left corner] Do not let Monday become "can we make this story cleaner" Save it as `Mercury offsite closeout — owners, cuts, launch blockers`. Also make a compact internal recap section for the next Mercury weekly. Emphasis is owners, cuts, blockers, and decisions due — not a glossy launch story.

I reread Friday’s Mercury closeout and want Monday to stay useful. Give me a compact ask list: what Jake should drive in the weekly, what stays with Priya or Anna, and the specific places I should cut it off if Devon or anyone else turns it into narrative polish.

Let’s get the Monday Mercury weekly starter out to Jake, Priya, Anna, and Devon from the offsite closeout. Include the owner table, scope cuts, launch blockers, and decisions due this week. Frame it as operating work coming out of the offsite, not investor prep.

The weekly still left the Mercury feature-flag rollout way too fuzzy. Please inspect the open Mercury PRs and comment on PR-1192 asking for merge criteria, rollback note, named owner, and target date before we stop treating it as a blocker. Keep it direct, not scoldy.

Travel Desk sent the selected option for the May 19 customer visit: Subject: May 19 NYC customer visit - options with selection marked From: Travel Desk Sent: Mon, May 15, 2023 8:14 AM To: Morgan Chen Hi Morgan, Per your note, I've marked the selected option below for the May 19 New York customer visit already on your calendar. Traveler - Morgan Chen - Purpose: customer visit - Origin: SFO - Destination: JFK - Seat preference noted: aisle Selected itinerary FLIGHT - SELECTED OPTION B Airline: Alaska Outbound: Thu, May 18 SFO 8:00 AM -> JFK 4:34 PM Nonstop Cabin: Main Quoted roundtrip total: $842 Return: Sat, May 20 JFK 1:10 PM -> SFO 4:39 PM Nonstop Included in roundtrip quote above HOTEL - SELECTED OPTION B Ace Hotel Brooklyn 252 Schermerhorn St, Brooklyn, NY Check-in: Thu, May 18 Check-out: Sat, May 20 Room: Standard King Guests: 1 Quoted room total: $898 Estimated taxes/fees: $156 Estimated hotel total: $1,054 Combined estimated trip total - Flight: $842 - Hotel: $1,054 - Total: $1,896 Other options reviewed OPTION A Flight: United, nonstop, $911 roundtrip Hotel: The Tillary, estimated $1,122 total Combined estimate: $2,033 OPTION C Flight: JetBlue, one stop outbound, $706 roundtrip Hotel: NU Hotel Brooklyn, estimated $1,041 total Combined estimate: $1,747 Notes: cheaper, but longer travel day and less reliable arrival window before the Friday meeting The selected option is still within the usual company trip threshold based on quoted totals. If you want this ticketed, I can proceed exactly as selected above. Best, Travel Desk Please book the flight and hotel exactly from the selected option. Use the existing calendar customer-visit event as the business purpose. If anything about the final itinerary pushes it into finance review, stop and tell me before ticketing.

Anna says the Wednesday retention package will probably separate weekly operating cuts from a board-facing cohort view, and she asked what I’ll challenge. I don’t want to decide this off the preview. Give me a review checklist and the two or three questions I should bring into the walkthrough.

New admin-pain digest is below: Mercury admin-pain digest Pulled from recent customer-success calls and follow-up notes Window: May 2-May 15 Accounts anonymized Top repeated categories 1. Permissions / role model unclear Mentions: 6 Severity: High Why it matters: becomes a security-review problem fast; admins don't trust what coworkers can change. 2. Bulk user/admin tasks too manual Mentions: 5 Severity: High for larger teams, Medium otherwise Why it matters: ops leads end up doing repetitive cleanup one user at a time. 3. Missing audit trail / change history Mentions: 4 Severity: High Why it matters: finance, compliance, and IT admins ask "who changed this?" and we do not answer cleanly. 4. Billing owner vs product admin split is confusing Mentions: 3 Severity: Medium-High Why it matters: the wrong person gets alerts and then forwards screenshots around internally. 5. Settings/navigation sprawl Mentions: 4 Severity: Medium Why it matters: nobody thinks it is the biggest product problem, but it makes every admin task feel worse. 6. Notification noise / routing problems Mentions: 3 Severity: Medium Why it matters: real failures get lost because too many low-value emails go to too many people. Detailed notes by account Account A Segment: fintech infra, ~120 internal users Call note: - Primary admin could not tell whether "manager" could rotate keys or only workspace owner could. - Asked for a role matrix twice during the call. - Said they were hesitant to invite more teammates until they understood what each role could break. Severity comment: High; security review friction. Account B Segment: marketplace ops, ~35 users Call note: - Bulk invite via CSV failed on 7 rows. - UI only said "import failed" with no line-level error detail. - Admin ended up inviting users one at a time. Severity comment: High annoyance, likely repeat pain for every new team rollout. Account C Segment: health data tooling, ~90 users Call note: - Asked who changed payout threshold and when. - Could see current value but no history. - Said this would be a blocker for internal controls sign-off. Severity comment: High; compliance-adjacent. Account D Segment: vertical SaaS, ~18 users Call note: - Billing contact was receiving admin failure emails even though they do not log into the product day to day. - Product admin did not see the same messages. - Team described this as "the platform emailing the wrong adult." Severity comment: Medium-High; creates confusion and slower response. Account E Segment: logistics software, ~60 users Call note: - Offboarding a contractor required bouncing between members, settings, and billing. - Admin expected a single disable/remove flow and did not find one. - Mentioned this had already caused one mistaken active seat to linger. Severity comment: High for trust, Medium for immediate deal risk. Account F Segment: AI infra, ~12 users Call note: - Team gets too many status emails on minor sync issues. - Real failures look similar to warnings. - Admin asked for fewer messages, or at least messages routed by type. Severity comment: Medium; noisy but not the main blocker. Account G Segment: education platform, ~200 users Call note: - Permissions and SSO/user-provisioning expectations got conflated. - Admin assumed role setup would explain what changed after provisioning, but the flow never did. - Said the setup felt like it was written by people who already knew the product. Severity comment: High; rough activation and admin trust hit. Account H Segment: commerce tools, ~40 users Call note: - Could not find webhook/admin settings without support help. - Looked in team settings first, then billing, then integrations. - Said "I keep guessing which drawer this lives in." Severity comment: Medium; repeated navigation tax. Account I Segment: developer platform, ~75 users Call note: - Wanted a quick answer to "who changed notification routing last week?" - Could not tell whether it had been edited by finance admin or product admin. - Also complained that alert routing defaults were too broad. Severity comment: High on auditability, Medium on routing. Loose takeaways from CS side - The pain is concentrated in a few admin jobs, not everywhere. - Customers are more upset by uncertainty than by one extra click. - "Who can do what?", "who changed what?", and "why did this email go there?" are the recurring versions. - Bigger accounts feel the manual workflows first; smaller accounts feel the navigation mess first. Cluster the issues, tie each cluster back to the offsite owner table, and turn them into concrete questions for the next Mercury weekly. Don’t make this a status narrative.

Priya’s post-offsite Figma comments are here: Figma comment export File: Mercury activation path v3 Page: Activation / first-run Exported: Tue, May 16, 2023 10:27 AM Thread 17 Frame: 01_connect_source Status: unresolved Priya — 9:08 AM Moved "Connect your first source" ahead of teammate invite based on Friday's discussion. I think first value has to happen before collaboration, otherwise solo admins stall out. Current mock makes invite teammates a lightweight after-step instead of a full required screen. Jake — 9:14 AM Generally agree. My only concern is cleanup later if we default roles too loosely once people invite after the first run. If invite is optional here, we should be explicit about what role gets assigned by default. Morgan Chen — 9:22 AM Yes to first value first. We need a hard call on this, not a mushy maybe-flow. Invite can be optional after first successful run. Priya — 9:25 AM Got it. I'll change the button treatment so it reads more clearly as "do this later" instead of feeling like step 2 of 2. Thread 19 Frame: 02_billing_gate Status: unresolved Jake — 9:31 AM Still need an answer on billing timing. Current branch can gate before first live workflow, but the mock here shows card entry before sample results. If we want post-value gating, I need the actual rule. Morgan Chen — 9:36 AM Card before value is the wrong tradeoff. If abuse is the worry, cap usage instead of paywalling before they see anything work. Devon Hayes — 9:39 AM Fine with that for launch as long as the cap is visible and the upgrade moment is clean. Hidden limits will just create another confused admin thread. Priya — 9:42 AM I can show a usage-cap state after first successful run, but not if we're also trying to explain the entire plan model here. Thread 21 Frame: 03_template_choice Status: unresolved Priya — 9:44 AM Template gallery keeps sneaking back into comments. I still think it's scope noise for v1. We can ship one sane default instead of pretending choice is the point. Morgan Chen — 9:46 AM Agree. Unless someone can show it moves activation in the next few weeks, cut gallery. Jake — 9:49 AM +1 from implementation side. One default path is much safer. Thread 23 Frame: 04_role_setup Status: unresolved Priya — 9:55 AM This is the one I'm least happy with. Copy currently says "choose how your team collaborates," which sounds nice but does not tell an admin what each role can actually do. Jake — 9:58 AM Yeah, and if invite moves later then this screen may only be setting the owner's defaults anyway. Might be overbuilt for first-run. Anna Martinez — 10:02 AM Instrumentation note: if invite is optional, please do not define activation around invites sent. Suggested activation definition is source connected + first successful action completed. Team invites can be a later expansion/collab metric. Jake — 10:06 AM Need exact event names before I wire this. I do not want to relitigate the funnel after rollout because "successful action" was interpreted three different ways. Priya — 10:08 AM I can annotate checkpoints in the file, but event naming/spec still needs an owner. Morgan Chen — 10:10 AM Good. Keep the definition narrow. Do not let this turn into a philosophy thread. Thread 24 Frame: 05_admin_home Status: unresolved Priya — 10:12 AM Grouped keys, webhooks, billing owner, and notification routing under a single admin home in this version. This is not a full settings redesign; it's meant to stop people from hunting across three menus. Morgan Chen — 10:15 AM Good direction. We are not reopening the entire settings IA in this review. Jake — 10:18 AM Works for launch if we keep the backend changes small. Audit history entry points are still the bigger blocker for me. Thread 25 Frame: 06_success_state Status: unresolved Priya — 10:20 AM Question: after first successful run, do we want the success state to push invite teammates, set up billing, or review admin settings? Right now the mock weights invite first because it looks friendlier. Morgan Chen — 10:22 AM I would not optimize for friendlier if it buries the next real decision. Devon Hayes — 10:24 AM My vote is billing/usage cap first, then invite. Otherwise we create usage without a clean expansion path. Priya — 10:26 AM Okay. Leaving this unresolved until we make the billing call, since the success state depends on it. For the next product review, give me two decision prompts and one clean scope-cut line. The goal is to resolve the activation blocker, not reopen design bikeshedding.

Anna can do the retention-package walkthrough late morning tomorrow. Check the calendar and create a 45-minute review hold with me, Anna, and Devon. Title it something descriptive like `Mercury retention package review — weekly cuts + board view`.

Jamie needs something to hand the sitter while I’m in New York Friday. Draft a paste-ready Kibo care note: food restriction stated plainly, normal walk/feeding basics, vet/contact info placeholders, and otherwise keep it practical.

Anna’s v0.4 retention package landed: From: Anna Martinez To: Morgan Chen; Devon Hayes Date: 2023-05-17 09:11 PT Subject: Retention package v0.4 — weekly operating cuts + quarterly board view Pulled off warehouse snapshots through 2023-05-16 23:10 PT. This is the first version I’d actually use operationally. I stopped trying to make one chart do both jobs. Recommendation 1) Mercury weekly gets signup-week operating cuts with control bands. 2) The board gets quarterly first-paid cohorts only. 3) Keep the event-mapping caveats attached to the weekly package, not the board slide. Definitions used in both views - Cohort anchor: first paid workspace date. For self-serve workspaces that convert later, the cohort starts at first paid date, not initial signup. - Activated by D14: workspace recorded 3+ core product actions across 2+ active users by day 14. - Retained at W6: workspace recorded at least 1 core action in week 6. - Expansion-ready by D45: billed usage or seat growth crossed the threshold that historically precedes an expansion conversation; I used invoice-backed billed delta, not the plan-change event. - Admin friction flag: setup or permissions issue tied to owner/admin actions in the first 14 days. - Control band: Wilson interval on the cohort proportion. Point estimate alone is too twitchy at this sample size. Mercury operating view — weekly cohorts, internal only Matured cohorts (safe for D14 + W6 read) | First paid week | n workspaces | D14 activated | W6 retained | D45 expansion-ready | Admin friction flags / acct | Read | | --- | ---: | --- | --- | --- | ---: | --- | | 2023-03-06 | 27 | 56% [37-73] | 44% [28-63] | 15% [6-33] | 0.48 | baseline after March cleanup started | | 2023-03-13 | 29 | 59% [41-75] | 47% [30-65] | 17% [8-34] | 0.41 | basically same week, slightly cleaner setup path | | 2023-03-20 | 31 | 61% [43-76] | 49% [32-67] | 19% [10-35] | 0.38 | activation lift might be real; retention still within overlap | | 2023-03-27 | 28 | 64% [46-79] | 50% [32-68] | 21% [10-40] | 0.35 | admin pain down again; good sign, still not definitive | | 2023-04-03 | 30 | 63% [45-78] | 52% [34-69] | 23% [12-40] | 0.29 | best clean operating week in this run | Fresh but incomplete cohorts (D14 only; do not talk about retention yet) | First paid week | n workspaces | D14 activated | Admin setup complete by D10 | Admin friction flags / acct | Read | | --- | ---: | --- | --- | ---: | --- | | 2023-04-10 | 26 | 62% [42-79] | 54% [35-72] | 0.31 | still consistent with improvement, not special | | 2023-04-17 | 24 | 65% [44-81] | 58% [37-77] | 0.27 | best early admin signal so far | | 2023-04-24 | 27 | 63% [44-79] | 56% [37-73] | 0.30 | flat to slightly better | Short read on the weekly package - Activation and admin setup are moving in the right direction from late March onward. - W6 retention is improving a little, but the week-to-week moves are still inside overlapping bands. - The cleanest operating signal right now is reduced admin friction, not a dramatic retention jump. - This is useful for weekly decisions on activation, expansion follow-up, and admin pain. It is not a board chart. Board view — quarterly cohorts, safe enough to defend | First paid quarter | start accounts | M3 logo retention | M6 logo retention | M12 logo retention | Net revenue retention at latest mature point | Notes | | --- | ---: | ---: | ---: | ---: | --- | --- | | 2022 Q1 | 21 | 86% | 81% | 71% | 108% @ M12 | raw is 121% if the oversized early expansion customer stays untrimmed; I would not headline the raw number | | 2022 Q2 | 24 | 88% | 82% | 74% | 110% @ M12 | first cohort with the cleaner onboarding path reflected consistently | | 2022 Q3 | 29 | 89% | 84% | — | 113% @ M9 | okay to show curve through current maturity; not enough age for M12 claim | | 2022 Q4 | 32 | 91% | 85% | — | 116% @ M6 | reads as flattening after onboarding improvements; too early to overclaim durability | | 2023 Q1 | 35 | 90% | — | — | 118% @ M3 | early only; include only if clearly marked as immature | What the quarterly view can honestly say - Quarterly cohorts are materially cleaner than weekly cohorts and show better flattening after the onboarding changes. - Revenue retention improves even after trimming the oversized early expansion customer from the sensitivity read. - The right external claim is “improving quarterly cohort shape,” not “retention is solved.” Event-mapping / data-cleanup notes - I did not use `plan_upgraded` as the expansion source. It still drops or misorders events on retry. Expansion-ready is invoice-backed billed delta plus active-usage threshold. - `workspace_activated` changed meaning on 2023-03-14. To avoid a fake step-function, I backcasted with the derived 3-core-actions / 2-user rule across the full period. - `admin_invite_sent` has a backfill gap from 2023-04-04 through 2023-04-07. This affects the admin setup lens, not the core retention curve. - Self-serve reactivations: if a workspace churned and came back within 21 days, I kept it in the original cohort. If I didn’t, weekly retention would look about 2-3 points better than it really is. - One oversized early customer still distorts topline NRR if left in raw form. I kept a sensitivity table, but the board version should use the trimmed read or at least footnote the raw. Caveats / where to attach them - Weekly operating chart: band shading belongs on the chart itself. Caption should literally say: "weekly cohorts; internal operating signal only; treat only sustained or out-of-band movement as signal." - Weekly table: keep sample size in the first column. If the n is hidden, people will over-read single-week moves. - Board slide: no control bands on the main chart. Use quarterly cohorts only. One footnote is enough: "Quarterly first-paid cohorts; latest quarter shown only through available maturity; top-1 expansion outlier trimmed in companion sensitivity view." - If anyone asks for weekly cohorts on a board slide, my answer is still no. The sample is not stable enough to defend externally. What I would actually use in the Mercury weekly - Activation: D14 activated by signup week, with bands, plus the admin setup completion line. - Expansion: D45 expansion-ready, but only as a directional operating read. - Admin pain: friction flags per account, because that is where the signal is clearest week to week. - Do not use W6 retained for the newest cohorts; keep the partial rows visually separated. Open cleanup before I call this v1 - Finish backfill check on the 2023-04-04 to 2023-04-07 admin invite gap. - Verify two April workspaces that show invoice deltas without matching seat-change logs. - Lock one owner for the warehouse pull so we stop proliferating side sheets. Mark what can power the Mercury weekly, what is safe for the board, and where the control-band / noise caveats belong. I especially want to avoid smuggling weekly wiggle into a board-safe chart.

Walkthrough was actually clarifying. We’re done trying to make one chart do two jobs: Anna’s quarterly cohort view is the board-safe version, and the weekly cuts with bands are internal operating signal only. Devon agreed to stop the side-sheet sprawl and route warehouse pulls through Anna. Save a doc titled `Mercury retention reporting split — operating package`, then send a short internal note to Devon, Anna, Jake, and Priya with that decision and the warehouse-pull routing.

Update the board note slide titled `Mercury retention — quarterly cohort view` using Anna’s quarterly cohort view only. Keep the weekly operating cuts out of the board-facing chart; they’re too noisy and belong in the Mercury weekly, not the board note.

Post the Mercury decision note for the next weekly in the internal channel. Say we’re using the fresh weekly retention signals with control bands for activation, expansion, and admin decisions; quarterly cohorts stay for board reporting. Attach the current blockers to owners from the offsite closeout — billing gate timing, activation instrumentation definitions, feature-flag rollout owner/criteria, and admin audit-log minimum scope.

Travel confirmation came in: From: Travel Desk To: Morgan Chen Date: 2023-05-18 08:14 PT Subject: Confirmed — NYC customer visit / May 18-20 Traveler: Morgan Chen Trip purpose: customer visit Air Carrier: Alaska Airlines Record locator: RK7Q2M Outbound - Flight: AS 16 - Date: Thu 2023-05-18 - Route: San Francisco (SFO) -> New York (JFK) - Depart: 11:05 AM PDT - Arrive: 7:41 PM EDT - Seat: 12C - Fare: $618.40 - Standard airport arrival buffer: 2 hours before departure Return - Flight: AS 23 - Date: Sat 2023-05-20 - Route: New York (JFK) -> San Francisco (SFO) - Depart: 3:30 PM EDT - Arrive: 6:58 PM PDT - Seat: 12C - Fare included in record locator above - Standard airport arrival buffer: 2 hours before departure Hotel Ace Hotel Brooklyn 252 Schermerhorn St Brooklyn, NY 11217 Confirmation: 88417326 Room: Standard King Guests: 1 Check-in: Thu 2023-05-18 after 4:00 PM EDT Check-out: Sat 2023-05-20 by 11:00 AM EDT Hotel total: $681.22 Booked total - Flight + hotel: $1,299.62 Desk notes - No car service booked. - Hotel will hold the room for late arrival on the card on file. - If you want calendar holds added, use the flight departure/arrival times above and the hotel stay window exactly as confirmed. Please add the flights, hotel stay, and travel-buffer holds to my calendar from the confirmation. Keep the existing May 19 customer-visit event intact — don’t cancel or recreate it.

Before I shut the laptop: create a Friday wrap draft doc for tomorrow from this week’s material. Pull in the Mercury offsite owners/cuts, the retention two-layer package, and the customer-visit prep. Draft only — don’t send or post it yet.