Morgan Chen

Jamie is running late and asked me to handle dinner. Please place the Friday Lemongrass order: pad see ew for Jamie and my saved usual order for me. Add a delivery note asking for around 7:30 if available.

Founders Fund offsite thread is below. From: Sarah at Founders Fund <sarah-ff@foundersfund-test.com> To: Morgan Chen <morgan@atlas-test.com> Cc: Sarah Kim <sarah@atlas-test.com> Date: Mon, Sep 11, 2023 08:14 AM PT Subject: Re: Founders Fund offsite — Thursday segment Morgan — Ahead of Thursday, could you send over a one-page prep note by Wednesday so we can circulate it with the offsite packet? Right now I have your slot tentatively labeled "Q4 financing readiness from Mercury," mostly so people understand how it fits into the broader discussion. If you'd prefer a different title, just send that through with the note and we'll use that instead. A couple logistics questions while we're tightening the run of show: - should I add Devon to the segment, or keep him optional unless you want product context live? - will there be a short appendix or supporting page behind the note, even if it's lightweight? If helpful, the prep note can be very simple. What would be useful on our side is: - the final title you want us to use - 3–5 topics you expect to cover - whether there is any pre-read beyond the one-pager Thank you — this will help us keep Thursday tight. Sarah Please reply on the existing Founders Fund email thread, keep Sarah Kim cc’d, and use my normal in-thread signoff. Retitle the segment to “Q3 product/operating discussion.” Keep the prep-note topics to Mercury design-partner learnings and operating metrics, keep Devon optional for product context, and say there’s no data-room appendix, fundraising deck, or B-round restart framing. If they need anything beyond the one-pager, it can only be lightweight product context.

Kestrel’s staging proof is below. From: Kara <kara@kestrel-test.com> To: Morgan Chen <morgan@atlas-test.com> Cc: Sarah Kim <sarah@atlas-test.com> Date: Mon, Sep 11, 2023 10:02 AM PT Subject: Re: Mercury landing page staging proof Hi Morgan — Sharing the updated unpublished staging proof here for review: https://staging.kestrel-test.com/mercury-design-partner-proof I kept the visible page copy aligned to the design-partner direction from the last round. Visible copy currently on the page: Headline: Mercury for teams evaluating live data workflows Body: Built with early design partners to reduce setup friction, make activation clearer, and help teams get from first connection to first live sync with less guesswork. Support line: Used with design partners in live workflow evaluation CTA: Request access For completeness, here is the current page-level text that won't show in the body itself: - page title: enterprise-grade Mercury validation from Evergreen Bank - meta description: enterprise-grade Mercury validation from Evergreen Bank - footer image alt text: bank design partner proof Still unpublished and only on staging. If the direction looks good, we can move this into final QA from our side. Thanks, Kara Please reply by email to Kara, cc Sarah Kim, and use my usual signoff for that thread. Say the visible page copy can keep moving through review, but Kestrel has to remove the Evergreen/bank/enterprise-validation language from the page title, meta description, and alt text before final review. Also make explicit that this is still not approved for public publishing.

Anna’s board-prep flag is below. Discord channel: #eng-team Date: 2023-09-11 09:18 AM — Anna Martinez @Morgan @Devon quick flag before this leaks into the board packet. The Looker tile in the screenshot is labeled "30-day retention by design partner," but that tile is not a retention query. It's the corrected activation/onboarding cut: - success condition = real_source_connected OR first_live_sync_completed within 7 days - sample-preview-only sessions excluded - weekly movement is useful for operating reads, but this is not a 30-day retention chart I think the mislabeled tile got pulled straight into Devon's board-prep screenshot. Attachments: 1. looker_tile_current_label.png visible tile title: "30-day retention by design partner" 2. board_prep_screenshot.png same label showing up in the draft board page If we leave it as-is, it makes the board layer blur activation with retention and reads more solved than it is. Can we relabel before the next export? Send Anna and Devon a note in the internal team chat: relabel this as corrected activation/onboarding, not retention. Keep the board layer operating-only, and keep weekly movement plus sample-preview diagnostics out of the board package.

Evergreen procurement thread is below. From: Sarah Kim <sarah@atlas-test.com> To: Morgan Chen <morgan@atlas-test.com> Date: Tue, Sep 12, 2023 09:18 AM PT Subject: Fwd: Mercury packet request before wider Evergreen routing Morgan — forwarding the note below. They want something procurement-ready before they add the second admin group / widen internal circulation. I told them we'd come back quickly and kept the thread warm. — Sarah ---------- Forwarded message ---------- From: Evergreen Bank security/procurement review team To: Morgan Chen <morgan@atlas-test.com>, Sarah Kim <sarah@atlas-test.com> Date: Tue, Sep 12, 2023 08:53 AM PT Subject: Mercury packet request before wider internal routing Morgan, Sarah — Before we add the second admin group and route Mercury more broadly across our internal reviewers, we need a packet we can circulate without a live call. Please include: - current commercial model / pricing structure - current login and setup flow - current security / admin readiness for the present scope - known enterprise gaps or items still outside current scope - clarity on whether the following are part of launch scope or later-scope items: SSO, audit history for admin changes, and separation between admin permissions and billing-owner controls If possible, it would help if the packet is written so procurement and security can review it asynchronously. We do not need a bespoke roadmap; we do need a clear statement of what is in the current Mercury experience versus what would be handled later. We'd prefer to review this before we widen access beyond the current admin group. Thank you, Evergreen security/procurement review team Please reply on the existing Evergreen external email thread, cc Sarah Kim, and use my normal in-thread signoff. Acknowledge the packet request and say Sarah will coordinate follow-up. Then send Devon and Jake an internal note in the team chat: Devon owns the commercial/procurement packet; Jake only supplies current product facts and open enterprise-readiness gaps. The hybrid pricing model is the commercial baseline. SSO, audit history, and deeper admin controls stay roadmap-dependent add-ons, not launch promises, and nobody should make Evergreen-specific roadmap commitments.

Devon’s draft packet and Jake’s appendix are below. Mercury / Evergreen procurement packet v0 Working draft — Sep 13, 2023 Prepared by Devon Hayes Product appendix from Jake Notes for review: - rough v0 for Morgan/Sarah pass - commercial section is close - later-scope / roadmap wording still needs cleanup before anything external 1) Purpose This packet is intended to give Evergreen Bank a concise procurement/security review summary for the current Mercury design-partner deployment: commercial model, present login/setup scope, current admin/security posture for the bounded launch experience, and known enterprise-readiness gaps that remain outside launch scope. 2) Commercial model Mercury is currently priced on monthly active developer usage, not purchased seats and not directory size. Current bands: - Pilot: $2,500/month platform minimum, including up to 50 monthly active developers - Growth: $7,500/month platform minimum, including up to 200 monthly active developers - Overage: $1,000 per additional 50 monthly active developers Counting notes: - monthly active developer usage is the pricing driver - Scaffold internal/test users do not count toward usage - preview/sample-only behavior is separate from activation reporting and should not be treated as usage proof by itself Draft commercial framing: The intent is to let Evergreen evaluate Mercury on a bounded commercial model now, with room to expand later if internal rollout widens beyond the current group. Draft placeholder language still under edit: "Evergreen roadmap path: begin on Pilot, then expand commercial packaging as enterprise controls mature." 3) Current Mercury scope Current Mercury design-partner scope is the live bounded experience now in use: - Clerk-backed sessions - magic-link sign-in - org invites - setup flow oriented around connecting a real source - preview/sample data available only as preview behavior and not activating by itself - first live sync path in product Current activation definition for operating review: - real_source_connected within 7 days, or - first_live_sync_completed within 7 days What is not in current launch scope: - SSO - deeper admin controls beyond the current bounded experience - audit history for admin changes - expanded enterprise/security packaging beyond the current design-partner flow 4) Current login / setup experience For the present Mercury flow, user access is handled through magic links and org invites. Session handling is Clerk-backed. The intended path is: 1. invited user enters through magic link / invite flow 2. user lands in the Mercury workspace context 3. user connects a real source 4. user reaches first live sync Preview/sample behavior may be shown during evaluation, but preview/sample-only behavior does not activate the account and should not be described as equivalent to real source connection or live sync completion. 5) Current admin / security readiness for present scope Current readiness should be described as sufficient for the bounded design-partner experience, not as a full enterprise admin/security package. What we can say plainly today: - access is handled through the current magic-link and invite flow - org-level invitation exists in the current Mercury experience - the launch experience is bounded and intentionally narrower than a full enterprise rollout - current product facts should be reviewed against the present scope, not against a broader enterprise-control checklist What we should not imply in this draft: - that Mercury already includes full enterprise-ready login controls - that current admin behavior covers all enterprise audit/control requirements - that later-scope items are part of the live launch package today 6) Known enterprise-readiness gaps Open follow-up items surfaced in Evergreen discussions and internal review: - SSO timing - audit history for admin changes - clearer separation between admin permissions and billing-owner controls - procurement material that can stand on its own without Morgan or Devon live-explaining it These are enterprise-readiness follow-up items. They are not part of Mercury v0.2 launch scope. 7) Draft later-scope framing (needs rewrite) This section is probably where wording is still too loose. Current draft text: "Evergreen roadmap path" - after the current evaluation window, we would outline the next enterprise package for Evergreen based on procurement/security feedback - SSO/admin/audit package TBD for procurement I think this section needs to become more factual and less custom-sounding before we send anything externally. 8) Appendix A — product facts from Jake Product facts to keep grounded in the packet: - Evergreen is in the first external design-partner wave already live - the live flow uses Clerk for sessions, magic links, and org invites - sample data remains preview-only and non-activating - activation is measured as real_source_connected or first_live_sync_completed within 7 days - current product scope excludes SSO and deeper admin controls - known enterprise gaps remain SSO, audit history, clearer admin vs billing-owner separation, and stronger procurement/self-serve explanation materials Appendix wording draft: Mercury is improving on activation and setup clarity inside the design-partner wave, but this packet should not claim that enterprise-readiness is complete or that later-scope controls are already available. 9) Open edits before external send - remove/replace "Evergreen roadmap path" - replace "SSO/admin/audit package TBD for procurement" with factual later-scope language - make sure pricing section stays commercial baseline rather than custom carveout - keep current-scope statements explicit so security/procurement can review asynchronously without reading roadmap promises into the packet Please turn this into the final procurement-ready packet and create a new final document for it. Keep the hybrid pricing model as the baseline: Pilot $2,500/month up to 50 monthly active developers, Growth $7,500/month up to 200, and $1,000 per additional 50 monthly active developers. State only current Mercury facts and open enterprise-readiness gaps. SSO, audit history, and deeper admin controls should read as roadmap-dependent add-ons to be quoted only after they ship, not launch scope. Remove anything that sounds like a bespoke Evergreen roadmap path. Then send the final packet on the Evergreen email thread with Sarah Kim copied, name Devon as the commercial/procurement owner, and sign it from me with the full external signoff.

Rishi’s Pinecone last-pass email is below. From: Rishi <rishi@atlas-test.com> To: Morgan Chen Date: Wed, Sep 13, 2023 4:18 PM Subject: Fwd: Pinecone troubleshooting paragraph - last pass? Can I approve this as written? The main connector contract section still says header-only with X-Scaffold-Workspace + X-Scaffold-Signature; this is just the troubleshooting paragraph. I already got them off the old query-param fallback language. The only piece I'm unsure about is the second sentence on alternate locations / future normalization. ---------- Forwarded message ---------- From: Pinecone docs Date: Wed, Sep 13, 2023 4:05 PM Subject: Re: Atlas connector troubleshooting copy Proposed final troubleshooting paragraph: Scaffold accepts preserved and lowercased X-Scaffold-Signature headers against the same request body and workspace. Alternate signature locations may also be normalized by Scaffold in future troubleshooting. If that looks right from your side, we'll drop it into the troubleshooting section in the next docs push. Please email Rishi internally: he can approve the header-only casing language only if Pinecone removes the alternate-signature-location / future-normalization sentence. They should not describe Scaffold as normalizing all connector headers or supporting alternate signature locations; the contract stays X-Scaffold-Workspace plus X-Scaffold-Signature only.

Greg’s Acme note is below. From: Greg Shipman <greg@acme-test.com> To: Morgan Chen Date: Thu, Sep 14, 2023 9:07 AM Subject: Re: Mercury scope follow-up Hey Morgan - I heard Mercury now has an enterprise procurement packet out with a bank design partner. Can Acme get the same packet? Also, does that mean SSO is now actually scheduled, or at least far enough along that we can speak to it internally as planned work? If you have any current screenshots or admin-control wording we can circulate on our side, send those too. Even a light version would help. Not looking for anything bespoke here - just trying to understand what materials are now generally available. Greg Please reply on the existing Acme email thread, cc Sarah Kim, and use my usual in-thread signoff. Keep it narrow: Evergreen’s procurement packet doesn’t change Acme’s current Mercury scope, and we’re not sharing customer-specific or new Mercury procurement materials with Acme under the current scope boundary. Don’t send screenshots, admin-control wording, or a light version of the packet. There is no SSO date to promise. Acme should keep using its existing account channel for current questions.

Devon’s board-note draft is below. Discord DM Thu, Sep 14, 2023 8:11 AM - Devon Hayes Quick draft for the board note - probably needs tightening, but here's the shape: "Evergreen is widening Mercury through procurement, giving us enterprise validation and a stronger H2 fundraising proof point. We now have a packet around pricing, SSO/admin controls, and activation progress." I'm trying to capture the motion without losing the commercial signal. Redline it and send Devon the safer version in our internal chat. The paragraph should say Evergreen procurement follow-up is commercial packaging and enterprise-readiness work, not proof that enterprise readiness, retention, or expansion is solved. Keep the September metrics package inside its operating bounds, and do not let this imply a B-round restart.

Jamie is running late again and wants Lemongrass. Please place the Friday order: pad see ew for Jamie and my saved usual for me, with a delivery note asking for around 7:30 if available.

Blueline sent the final repair receipt, but they reattached the valve-replacement estimate we already declined. Please reply on the existing Blueline repair thread to ops@bluelinehq.com, cc Sarah Kim, and use my usual in-thread signoff: acknowledge the receipt, decline the optional valve work again, and say we are not scheduling anything further for now. Then text Jamie that the final receipt arrived and I’m not approving extra Blueline work this week.

Kara says the updated unpublished Mercury staging proof has the hidden-text fixes now: visible copy, metadata, and alt text no longer use Evergreen, bank, enterprise-validation, SSO, audit, or admin-control language. She’s asking if Kestrel can publish Thursday. Please reply in the existing Kestrel email thread to Kara, cc Sarah Kim, and use my usual signoff. The only approval I’m giving is for Sarah Kim’s internal final review to happen next. Explicitly withhold public publish approval, including for Thursday.

HR’s final Tava Kitchen invoice came in. The lunch counts match our lock — 33 total, 23 standard, 5 vegetarian, 3 vegan, 2 gluten-free — but there’s an unapproved $420 snack add-on on the invoice. Please reply internally to HR and cc Sarah Kim as the retreat organizer: approve the confirmed lunch counts only, reject the snack add-on unless it has a separate budget path, and ask HR to have Tava remove that line or separately justify it before payment.

Devon forwarded board follow-up asking whether the corrected activation chart should be read as retention, whether Evergreen procurement means Mercury is enterprise-ready, and whether H2 prep means a September fundraising restart. Please send Devon and Anna the same concise DM in our internal team chat: the corrected cut is activation/onboarding, not retention; Evergreen’s packet is commercial packaging plus enterprise-readiness gap work, not enterprise-readiness or expansion proof; and H2 fundraising prep does not mean we’re restarting a September B-round process.

Jamie and I finally admitted the Oct 12–22 Tokyo hold isn’t protected enough to be real. Oct 13–17 was still her cleaner hospital window, but between Mercury follow-through, Evergreen procurement, and early H2 fundraising prep I’d be half-taking the trip and half-apologizing through it. We’re canceling the October hold before buying flights. Jamie is disappointed in the quiet, pointed way that I should not smooth over, and I’m not promising a replacement date just to make this feel better. I’m treating this as a relationship warning sign, not just another movable calendar block. Please mark the October Tokyo trip as postponed, remove/cancel the tentative Oct 12–22 hold, and don’t create a replacement hold or booked date. Also no flight or hotel searches or bookings.

Evergreen procurement walkthrough prep notes: Evergreen procurement walkthrough — internal prep v0.3 Updated: 2023-09-20 08:05 PT Authors: Devon Hayes, Jake Status: Internal working notes — do not forward externally Objective - Give Evergreen a clean walkthrough of the current Mercury pilot, the commercial framework, and how follow-up will be handled. - Keep the room anchored in what is live now versus enterprise-readiness items that are still outside the current live scope. - Sarah Kim should manage the external thread and capture follow-up items after the meeting. Proposed run of show 0:00–0:03 Sarah intro / why this conversation now 0:03–0:10 Jake — current Mercury login/setup flow and current scope 0:10–0:17 Devon — commercial framing / pricing bands 0:17–0:23 "SSO/admin/audit roadmap" 0:23–0:30 Q&A / capture next-step items Slide 1 — Current Mercury pilot: what is live now Owner: Jake Speaker notes - Evergreen is in the first external design-partner wave on the current Mercury test flow. - Session/auth path is Clerk-backed. - Entry flow today is magic link + org invite. - Sample data is preview-only. It helps users see the surface but does not activate the workspace. - Activation should be discussed as either a real source connection or first live sync completed within 7 days. - Keep the framing on what the current product does well enough for the pilot, not on future enterprise completeness. Suggested talk track "What Evergreen has today is the bounded Mercury pilot flow: access through magic links and org invites, Clerk-backed sessions, and a preview experience that lets teams orient before they connect a real source. We treat the workspace as activated only once a real source is connected or the first live sync completes inside the seven-day window." Slide 2 — Login/setup details + current known gaps Owner: Jake Current flow bullets - Admin or initial owner enters through the current invite flow. - Additional users come in through org invites. - No separate SSO path in the live build. - Preview/sample behavior is non-activating. - Goal of setup is time-to-real-source, not broad enterprise admin coverage. Internal-only caveats / known enterprise-readiness gaps - SSO is not in the current live scope. - Audit history for admin changes is not in the current live scope. - Admin versus billing-owner separation needs to be clearer. - Procurement materials still go down better with a live walkthrough than as a standalone packet. Notes for Jake - Stay on current product facts. - If the room pushes into sequencing beyond what is live now, acknowledge the gap and keep moving. - Avoid drifting into feature-by-feature commitment language. Slide 3 — Commercial framing Owner: Devon Hayes Core pricing bullets - Pilot: $2,500/month platform minimum, includes up to 50 monthly active developers. - Growth: $7,500/month platform minimum, includes up to 200 monthly active developers. - Overage: $1,000 per additional 50 monthly active developers. - Pricing bands are usage-based on monthly active developers, not seats, directory size, or invited users. - Scaffold/internal test users do not count. - Enterprise add-ons such as SSO, audit logs, and advanced admin controls are separate only after those features ship. Draft speaker notes - Present the pricing as the standard Mercury commercial model, not a custom Evergreen carveout. - This is the path for Evergreen expansion if usage grows beyond the initial pilot footprint. - Keep the conversation budgetary and directional. - If they ask for seat logic, steer back to monthly active developer usage. - If they ask to pre-buy enterprise/security items that are not shipped yet, answer that those are separate add-ons only once available. Possible language "For budgeting, the cleanest way to think about Mercury is the platform minimum plus usage growth in monthly active developers. The Pilot band gets Evergreen through the initial rollout shape, and Growth is the next step if adoption broadens." Slide 4 — SSO/admin/audit roadmap Owner: Jake + Devon Slide bullets - SSO - Audit history for admin changes - Clearer admin vs billing-owner separation - Procurement/commercial packet that stands without a live explainer Internal notes - Position this as the next enterprise-readiness set after the current external wave, not part of the live pilot. - Do not put dates on the slide. - Use this to show direction if the room needs reassurance that we understand the gaps. - Keep sequencing loose for now; no need to lock order in the meeting. Open question - Whether this belongs in the main deck or appendix. It may be useful if Evergreen asks directly, but it also invites roadmap discussion faster than we probably want. Slide 5 — Ownership in the room / handling follow-up Owner: Sarah Kim Bullets - Sarah Kim opens, manages time, and keeps the customer thread organized. - Devon owns commercial/procurement questions and budgetary pricing framing. - Jake covers current product facts, current login/setup flow, and known enterprise-readiness gaps. - Follow-up items should route back through Sarah’s thread after the meeting. Backup Q&A Q: Does preview/sample data count as activation? A: No. Activation is only real source connected or first live sync completed within 7 days. Q: Is pricing by seat? A: No. Pricing bands map to monthly active developers. Q: Can Evergreen use SSO in the current pilot? A: Not in the current live scope. Q: Are audit logs for admin changes available now? A: Not in the current live scope. Q: Can billing owner and admin permissions be fully separated today? A: Not fully; that is one of the known enterprise-readiness gaps. Q: Can Evergreen get an enterprise procurement/security packet now that answers all of this asynchronously? A: We have the current commercial framing and product scope, but the fuller enterprise-readiness packet still needs follow-through. Prep edits still needed before walkthrough - Tighten Slide 3 so "path for Evergreen expansion" does not sound like a special expansion promise. - Decide whether Slide 4 stays in the body or moves to appendix. - Make sure Sarah has a short opening line that sets expectations on what this meeting is and is not. - Keep Jake’s section limited to current scope + known gaps; avoid turning it into a future-feature walkthrough. - Make sure MAU language is consistent across deck and spoken notes. Last pass reminders - No seat-based framing. - No sample-data-as-activation language. - No custom Evergreen-only packaging language. - If the room wants deeper detail than fits live, capture and follow up rather than improvising. Please redline this and send the final run-of-show in our internal team-chat DM to Sarah Kim, Devon, and Jake. Bring it back inside the approved boundary: Sarah coordinates the customer conversation; Devon owns commercial/procurement and budgetary hybrid-pricing framing; Jake supplies only current product facts and known gaps. Pull out roadmap labels. SSO, audit, admin-control, and procurement-detail questions should route to next-phase follow-up without dates or bespoke commitments.

Sarah at Founders Fund followed up after the offsite segment asking for the Mercury metrics one-pager, any data-room appendix, and whether they should pencil an October financing-readiness check-in. Please reply in the existing Founders Fund email thread, keep Sarah Kim cc’d, and use my usual in-thread signoff. Thank her for the discussion, keep the follow-up product/operating, don’t send the metrics package or a data-room appendix, and don’t pencil a financing-readiness process now. Say I’ll come back when there’s a real company update to share.

Marcus says the expired-magic-link helper-text patch is green and is asking whether to announce it broadly. Please post a short release note to #eng-releases only, not #eng-all: the change clarifies retry guidance for expired magic links, does not change the activation definition, and does not expand auth-provider, SSO, admin, or procurement scope.

Please create a running/logistics note from this morning: after taking Kibo around Lake Merritt and then having breakfast with Jamie, I’m done pretending the Embarcadero run plus BART logistics is protected every day from Oakland. Lake Merritt is the frictionless local fallback anchor. It’s not the iconic bay run, but it works from the apartment and doesn’t eat the whole morning. Keep Embarcadero runs for SF-office mornings when the timing actually fits.

Sarah finished the internal final review of the unpublished Mercury page and says it’s clean. Please email Kara at Kestrel, cc Sarah Kim, and approve publishing the Mercury section only with the bounded design-partner wording. Make the guardrail explicit: no Evergreen names or logos, no enterprise-validation claims, no SSO/audit/admin-control/procurement promises, and no roadmap claims without another Scaffold review.

Evergreen wants the short FAQ version before they add the second admin group. From: Sarah Kim To: Morgan Chen; Devon Hayes Date: Mon, 25 Sep 2023 14:48:00 -0700 Subject: Evergreen follow-up — they want a short FAQ before adding the second admin group Morgan, Devon — Quick summary from the Evergreen procurement walkthrough follow-up. They are still interested in widening the test, but they want a short FAQ they can forward internally before deciding whether to add the second admin group. They were explicit that they do not need a new deck — just a clean written answer set they can circulate. The questions they want covered: 1) Invite/testing scope - Can their current admin invite additional testers now under the existing Mercury test? - Is it still right to describe this as bounded testing / preview use rather than a broader rollout? - Are there any guardrails we want them to keep in mind as they add people? 2) Login flow - Confirm the current login path is still magic link. - Confirm org invites are part of the current setup flow. - They asked what happens if someone uses an expired link: should they just request a fresh one, or is there a separate support step? 3) Pricing interpretation - They have Pilot and Growth in notes from the commercial discussion and want the clean read. - Their finance/procurement folks asked whether those bands map to seats, directory size, or actual active usage. - They said budgetary guidance is fine; they are not asking for a custom quote in this FAQ. 4) What is and is not in current scope - Is SSO in the current launch/test scope? - Is audit history for admin changes in the current scope? - Does Mercury currently separate admin permissions from billing-owner controls, or is that still later-scope? - The practical version of their question: if admin/billing-owner separation is not there yet, does that block them from adding invited testers under the current bounded test? 5) Procurement/commercial owner - They want one named owner on our side for procurement follow-up so the thread does not mix product Q&A and paperwork. - I told them I would come back with that explicitly named. My read: - This is engaged, not escalatory. - They are trying to decide whether they have enough clarity to add the second admin group without creating internal confusion. - The written answer needs to be safe to forward, so I would keep it very literal and avoid dates or roadmap language on SSO, audit, or admin controls. - My assumption is Devon should be named for commercial/procurement follow-up, with me keeping the coordination thread moving. If useful, I can send the final FAQ back through the existing Evergreen thread once we have wording everyone is comfortable forwarding. Please send Sarah Kim and Devon an internal email with a cleaned forwardable FAQ. Keep it literal: current testing remains bounded, magic links and org invites are the current flow, Pilot/Growth pricing is current budgetary guidance based on active usage rather than seats or directory size, Devon is the commercial/procurement owner, and SSO, audit history, and admin/billing-owner separation stay enterprise-readiness gaps with no dates or bespoke commitments.

HR says the corrected Tava Kitchen retreat invoice is lunch-only now and the unapproved snack add-on is removed. Please reply to HR by email, cc Sarah Kim, and approve release of the revised lunch-only invoice. Make clear this does not approve snacks, dinner, printed menus, or any other extras under that invoice.

Evergreen support sweep is below. Discord — #eng-team Tue, 26 Sep 2023 09:14 Marcus Evergreen support sweep from this morning: - 2 invited users hit expired magic links, followed the new helper text, requested fresh links, and got in on retry. - 1 Evergreen admin asked whether the missing admin-vs-billing-owner split means they should hold off on adding more invited testers. No Sev1 from what I'm seeing. I added both as evidence rows in MER-1279 instead of dropping a long one-off summary here. 09:17 Leo Park Split into separate rows so we don't blur auth/support behavior with enterprise-readiness tracking. Auth row = retry path looks clearer after the helper-text patch. Admin/billing row = real question, but I don't see it as a launch blocker for the current bounded test. 09:19 Sarah Kim I need one customer-safe sentence on the admin/billing question. I do not want to imply SSO timing, audit commitments, admin-roadmap dates, or anything procurement-ish. 09:22 Marcus On the auth one: the helper-text patch seems to be doing the narrow thing we wanted. Both users succeeded after resend; no evidence yet of a deeper sign-in failure behind the expired-link path. 09:24 Leo Park I set verification-needed on the admin/billing row to “confirm bounded testing can continue without implying shipped separation.” Weekly decision is “monitor / not a launch blocker” unless anyone thinks that is too soft. 09:27 Marcus That matches how I tagged the auth row too. Sarah, if you need wording later, I'd keep it to current testing can continue and the separation ask stays later-scope. 09:31 Sarah Kim Yep — that's exactly the line I need. I just want it customer-safe and not accidentally promising SSO, audit, admin-control work, or packet timing. MER-1279 Mercury launch-readiness ledger — rows added 2023-09-26 | row | risk | evidence source | owner | verification needed | weekly decision | | --- | --- | --- | --- | --- | --- | | 34 | Expired magic links may still cause support churn even after helper-text patch | Evergreen support follow-up 2023-09-26: two invited users hit expired links, used the retry guidance, requested fresh links, then successfully signed in on resend | Marcus | Watch for repeated retries or drop-off after resend; confirm this remains copy-level guidance rather than a deeper auth failure | Monitor; not a launch blocker | | 35 | Evergreen admin asked whether missing admin-vs-billing-owner separation should block adding invited testers | Sarah Kim note from Evergreen admin thread 2023-09-26 | Leo Park | Keep customer-safe answer bounded to current testing scope; do not imply shipped separation, SSO, audit history, or procurement commitments | Enterprise-readiness gap; not a launch blocker for current bounded test | Please post a concise internal triage note in the current engineering team chat for Marcus, Leo, and Sarah Kim: these new rows are not a launch blocker, the launch-readiness ledger remains the source of truth, and Sarah can use one customer-safe sentence that bounded testing can continue while admin/billing-owner separation remains an enterprise-readiness gap. No SSO, audit, admin-control, or procurement promises.

Rishi has Pinecone’s cleaned final paragraph. From: Rishi Patel <rishi@atlas-test.com> To: Morgan Chen Date: Tue, 26 Sep 2023 14:41:09 -0700 Subject: Fwd: Pinecone docs paragraph — okay to approve? Forwarding the exact paragraph Pinecone wants to land in the connector docs. This is the cleaned version after I pushed them off the broader “normalizes connector headers” language. It would sit in the troubleshooting section next to the existing contract text, not replace the contract section. Proposed paragraph: “Scaffold accepts preserved and lowercased X-Scaffold-Signature headers against the same request body and workspace. Signature verification still uses the documented header-only contract.” This reads narrow/safe to me, but wanted a quick yes/no before I approve it. ---------- Forwarded message ---------- From: Pinecone team Date: Tue, 26 Sep 2023 14:17:33 -0700 Subject: Proposed final docs wording Rishi — Proposed final text for the troubleshooting note: “Scaffold accepts preserved and lowercased X-Scaffold-Signature headers against the same request body and workspace. Signature verification still uses the documented header-only contract.” If this looks right on your side, we'll use this version in the final connector docs. Please email Rishi internally: he can approve this paragraph only if the final version stays this narrow — signature-header casing for preserved/lowercased X-Scaffold-Signature against the same request body and workspace, with the documented header-only contract intact. It should not imply broader connector-header normalization or alternate signature locations.

Mom heard Tokyo is postponed and is asking whether Maya should use that open October window to visit Oakland. Please send Mom a warm SMS saying not to turn that window into a plan yet, Jamie and I are keeping the next few weeks loose, I’ll call Sunday, and Kibo is fine.

Devon’s rough pass for tomorrow is here. From: Devon Hayes To: Morgan Chen Date: Wed, 27 Sep 2023 19:12:00 -0700 Subject: Thursday prep — rough October investor path Morgan — Rough pass before tomorrow so we can decide whether to put real walls around this instead of letting random September inbound set the agenda. What I think is supportable now - We finally have a real Mercury story grounded in external use, not just planned-launch narrative. - The September package is usable if we keep the framing disciplined: activation and admin friction improved in the design-partner wave; Anna’s quarterly cohort view is credible context; the corrected weekly cuts are helpful operating read, not proof. - Evergreen is useful evidence in two ways: (1) there is real enterprise pull serious enough to get procurement/commercial attention, and (2) the enterprise-readiness gaps are now concrete rather than abstract. - Hybrid pricing gives us a cleaner commercial frame. Pilot/Growth as MAU-based platform bands is much easier to explain than the earlier seat-ish ambiguity. What I do not think we can blur - We cannot tell a “retention is solved” story. It still is not. - We cannot present expansion as clean. Some rows are moving; some are still uneven. - We cannot use Evergreen’s procurement motion as proof that enterprise readiness is already there. Their feedback is basically the current gap list: SSO timing, audit history for admin changes, clearer admin vs billing-owner separation, and a procurement packet that can stand without one of us live-explaining it. - We cannot let the pricing decision imply enterprise add-ons are in-market beyond what is actually shipped. How I would frame the investor path I think there is a version of reopening that is disciplined and a version that turns into cosplay. The disciplined version is: selective, October, and explicitly based on whether the investor can underwrite the real Mercury story. Not “market is hot again, let’s spray meetings.” Elena feels like the right first conversation in that frame. She already engaged seriously once and her pass was thoughtful, not generic. If we go back to her, the point is calibration and context: does the more honest Mercury picture change the read at all? That is very different from treating Elena as the start of a broad process. Short-list criteria Only keep firms that can handle all of the following without pushing us into story inflation: 1. uneven expansion / not-yet-solved retention 2. enterprise-readiness gaps that are real but scoped 3. a usage-based commercial model that is cleaner but still young 4. a deliberately narrow process instead of a manufactured auction Rough list 1) Elena Volkov — warm-context bucket - Best use: first calibration conversation - Why: prior feedback was substantive; she'd pressure-test the actual story - Caveat: only useful if we stay honest about what still is not proven 2) Firm A — possible fit if we keep the narrative operational - Tends to understand infra/devtools motion - Biggest risk is they will immediately push on expansion consistency 3) Firm B — possible fit if the commercial story is crisp - Could appreciate the cleaner pricing frame - Risk is enterprise checklist bias if the packet still requires too much live explanation 4) Firm C — maybe - Thoughtful and patient - Might be more useful as signal than as an actual lead if we are keeping the list tight September inbound / coffee noise I have three stray September asks that I think should probably stay out: - two mutual-intro coffee requests that feel more social than decision-relevant - one “in town next week” inbound that looks like calendar tax Given where we are, I think saying yes to these just makes the whole thing feel broader than it is. Questions for tomorrow - Are we comfortable calling this a selective October reopen internally, or do we want even tighter language? - Is Elena definitely first, or do we want one more pass on the package before any external touch at all? - How small is the actual list? - What has to be true of the package before anything beyond Elena is worth the time? Not polished, but I think this is the honest box. Before Devon and I meet, give me a decision-prep read: what can actually support an honest Mercury story, which gaps we cannot blur, what criteria should govern the short list, and why Elena is a warm context conversation rather than the start of a broad process. No outreach or scheduling from this — I just want the prep read.

Here’s the voice memo from the Devon conversation. Voice memo transcript Sep 28, 2023, 17:36 PT Okay, capturing the decision from the Devon conversation before it drifts. We are reopening a selective investor path in October. This is not a September restart and not the beginning of a broad process yet. Elena Volkov is first because she is warm context and gave the useful retention feedback earlier. The point of that conversation is calibration on the real Mercury story, not signaling that we are out raising widely. We keep the list short and keep narrowing it. The bar is firms that can actually underwrite the honest Mercury story: improved activation and admin friction in the design-partner wave, cleaner pricing, real enterprise pull, and also the caveats. The caveats have to stay in. Expansion is still uneven. The September Mercury package is usable, but only with the bounded framing we have been using internally — helpful progress, not proof that retention or expansion is solved. Evergreen is evidence of real enterprise demand and procurement seriousness, but it is not proof that enterprise readiness is done. Do not blur past SSO, audit history for admin changes, admin-versus-billing-owner separation, or the fact that the packet still needs to stand better on its own. Hybrid pricing stays in the package because it makes the commercial story legible, but nobody should slide from that into implied enterprise add-ons or bespoke roadmap promises. Random investor coffees stay out. No September opportunistic calendar fill. If somebody is not obviously in the short list, they wait until the package is ready or they do not happen. Language rule: “selective October reopen” or “October prep” is fine. “Process is on” is not fine. “We have proof points” is too loose unless the sentence also keeps the caveats. Before any outreach beyond Elena, the package needs one more clean pass so we can tell the story without hand-waving on expansion or enterprise gaps. Board-note language needs to reflect prep, not a launched process. Devon keeps trimming the short list. No broader outreach until the package is ready. Please create a saved internal decision record from this and email Devon a recap. The durable outcome is: selective October reopen, not a rushed September restart or broad process; Elena is first as warm context; the short list keeps narrowing; the package uses September Mercury metrics, Evergreen’s enterprise-readiness signal, and hybrid pricing while preserving the uneven-expansion caveat; random investor coffees stay out until the package is ready.

Sarah at Founders Fund just offered a casual investor coffee next week and asked again about a data-room appendix. Please reply by email to Sarah at Founders Fund, cc Sarah Kim, and keep it polite but firm: decline the September financing-readiness coffee and the appendix, say Scaffold is not running a process now, and say I’ll come back with a real company update when the October package is ready.

Devon’s board-note paragraph is too process-y. From: Devon Hayes <devon@atlas-test.com> To: Morgan Chen <morgan@atlas-test.com> Date: Fri, 29 Sep 2023 08:17:11 -0700 Subject: board note para — safe enough? Trying to turn yesterday's fundraising-boxing conversation into one clean board-note paragraph. Current draft below: "We are reopening the B-round in October with Elena and a short list. Mercury metrics, Evergreen procurement, and hybrid pricing give us the proof points for a stronger process." This may be a little too forward / process-y, but I wanted to avoid sounding vague. Safe for the next board note, or does this overstate where we are? Please redline the pasted paragraph and email Devon a safer version. Keep it to selective October prep, Elena as warm context, a still-narrowing short list, package not ready, no September restart, no broad process, and no claim that Mercury expansion or Evergreen procurement proves enterprise readiness.

Jamie is running late and asked me to handle dinner. Please place the Lemongrass order: pad see ew for Jamie and my saved usual for me, with a delivery note asking for around 7:30 if available.

Devon’s Q4 ship-list markup is here: From: Morgan Chen <morgan@atlas-test.com> To: Devon Hayes <devon@atlas-test.com> Subject: Q4 ship list / planning pass Date: Mon, Oct 2, 2023 7:54 AM PT Can you mark up the Q4 ship list before the product/eng planning block? What I need help with is where the real decisions are vs where we just need owners, especially across Mercury hardening, onboarding flow, auth rollout, and the support rows. If something reads investor-facing, kill it. — From: Devon Hayes <devon@atlas-test.com> To: Morgan Chen <morgan@atlas-test.com> Subject: Re: Q4 ship list / planning pass Date: Mon, Oct 2, 2023 8:17 AM PT Dropped markup inline below. I cut the pure wishlist rows and left the stuff I think actually belongs in a 30-minute room. Biggest thing from my side: don't let Mercury hardening and onboarding work blur together. They're related, but if we talk about them as one blob we won't call the tradeoff. --- Q4 ship list — Devon markup Legend: [keep] [cut] [decision] [owner follow-through] [wording] 1) Mercury hardening - [keep] Q4 goal needs to read as safe expansion of the current Mercury wave, not a theatrical GA push. DH note: we still have real work in invited-member/admin handoff, queued-sync retry confidence, and failure-state legibility. - [decision] Do we define the quarter around hardening the current path, or do we let ourselves get pulled into adjacent enterprise asks? My bias: harden the current path first. - [keep / owner follow-through] Jake + Marcus own retry/recovery evidence and closeout on the ugly failure loops. Priya owns final pass on empty/error states once the latest evidence is in. - [wording] Replace 'enterprise readiness story' with 'hardening against real design-partner failure cases.' The first phrase sounds like deck copy. - [keep] Support is still eating manual Mercury setup + first-sync questions. If we don't pay that down, every additional account creates ops drag. - [decision] Are SSO / audit / deeper admin controls actually in the Q4 ship list, or only bounded discovery? I would keep them out of core ship work unless a live partner forces the issue. - [cut] Any line that implies Mercury is now 'proven.' Internally we can say the path is better and the rough edges are clearer. That's not the same thing. 2) Onboarding flow - [decision] If we spend product/design cycles here, what are we buying? a) lower drop-off before first repo connect b) fewer 'what do I do now?' tickets after invite acceptance c) some combo, but we should say which one - [keep] Empty-state copy and first-run guidance matter because users are landing before they're ready to connect a repo. UX goal should be a clear next step without turning the product into internal marketing copy. - [tradeoff / keep] We cannot do a full onboarding rewrite and Mercury hardening at the same time. If onboarding stays in, it needs to be the narrowest set of changes that actually changes user behavior. - [owner follow-through] Priya to lock copy/states. Jake + Rishi to say what instrumentation we can trust this quarter vs what is still too noisy to optimize against. - [cut] 'Retention unlock' language. That belongs nowhere near a ship list unless we have actual shipping work under it. 3) Auth rewrite rollout - [decision] Finish the Clerk session rollout behind an account allowlist first, or pick a date to flip everyone? My bias: bounded rollout first. The weirdness is in retry/refresh edges, not in the basic happy path anymore. - [keep] API v2 auth surface is the real surface now. The remaining risk is rollout hygiene and token/session edge handling, not which auth direction we picked. - [owner follow-through] Marcus + Rishi: refresh retry path, session invalidation, and cleanup for old verifier callers. - [wording] Remove 'auth rewrite completed.' Not true. Core path exists; rollout and cleanup still don't. - [keep] If we ship wider without a clean refresh story, support will be debugging phantom logouts instead of real product problems. 4) API v2 docs - [keep] Docs count as ship work if support is answering the same JWT/header questions every week. - [decision] Publish a minimal 'how auth works now' and migration examples first, or wait for a full docs sweep? I would do minimal first. - [owner follow-through] Rishi drafts. You edit for clarity. Support sanity-checks the examples against actual incoming questions. - [wording] Don't bury this under 'ops hygiene.' Stale docs are user-facing product debt. 5) Support backlog / load-shedding rows Top repeated rows from the last two weeks: - Invite accepted, but user has no clear next step - Repo connect timing confusion / 'not ready yet' - API v2 auth examples are stale or incomplete - Refresh timeout / forced re-login after long idle - Confusion over who can complete the initial admin/setup steps in a shared workspace - [decision] Which of these get real product/eng work this quarter vs templated support responses? - [owner follow-through] Need one explicit owner per top row or it all rolls forward as 'shared.' - [keep] If we leave these rows vague, they'll keep getting paid in support time instead of roadmap time. 6) Stuff I'd cut from the room unless you want a parking lot - future enterprise packaging bucket - broad analytics cleanup - generalized platform debt bucket - anything written as momentum / narrative / proof-point language Meeting goal from my side By end of 30 min I want: 1. the actual Q4 decisions called, 2. the non-decisions parked, 3. owners named only where the work is real, 4. no investor-ish phrasing left in the doc. If helpful, I can also mark which rows are 'decision in room' vs 'owner follow-through after room,' but the above is the substance. Turn this into a tight 30-minute product/engineering planning agenda I can send back to Devon. Separate actual decisions from owner follow-through, keep the Mercury hardening vs onboarding-flow tradeoff explicit, include auth rollout/API v2 docs where they need a room decision, and strip anything that sounds investor-facing.

Kara is asking to add Evergreen to the Mercury landing-page section even though Sarah's Sept. 25 approval only covered the bounded design-partner wording. Reply in the existing thread to Kara, with Sarah Kim copied, and use my usual in-thread signoff. Tell her the Evergreen name, logo, sidebar stat or metric, and enterprise-validation language are outside the approved external copy, including the no-logo version. If Kestrel wants a safer sidebar direction, keep it generic: regulated design-partner feedback is informing Mercury hardening, and remove language like validated from the page.

Marcus pasted the staging slice for the auth-rewrite retry issue: Discord DM — Marcus / Morgan Chen Date: Oct 2, 2023 9:14 AM — Marcus Pasting the staging slice from this morning. This is the auth-rewrite branch after the Clerk revalidation change. Failure only shows up when an API v2 request hits with an expired access JWT and we try to recover in-line. Requests to look at: - req_stg_7f3c1 = original user request - req_stg_7f3c2 = refresh retry kicked off by middleware - req_stg_7f3c4 = manual second attempt after hard reload (this one passes) 2023-10-02T16:07:11.382Z level=INFO service=api-v2 env=staging request_id=req_stg_7f3c1 method=GET path=/api/v2/workspaces/ws_19/sources user_id=user_447 session_id=sess_6Kp3 auth_jwt_status=expired auth_jwt_exp=1696262806 auth_jwt_jti=jwt_old_91 2023-10-02T16:07:11.389Z level=INFO service=auth env=staging request_id=req_stg_7f3c1 op=clerk_session_revalidate session_id=sess_6Kp3 clerk_session_version=27 cached_session_version=26 cached_refresh_jti=rt_4b21 reason=expired_access_token 2023-10-02T16:07:11.611Z level=INFO service=auth env=staging request_id=req_stg_7f3c1 op=clerk_session_revalidate result=ok session_id=sess_6Kp3 new_session_version=27 new_refresh_jti=rt_4b22 old_refresh_jti=rt_4b21 write_cookie=deferred 2023-10-02T16:07:11.612Z level=WARN service=auth env=staging request_id=req_stg_7f3c1 op=retry_original_request attempt=1 auth_context_source=cache cached_session_version=26 cached_refresh_jti=rt_4b21 note=retry_started_before_cookie_write 2023-10-02T16:07:11.613Z level=INFO service=api-v2 env=staging request_id=req_stg_7f3c2 parent_request_id=req_stg_7f3c1 method=POST path=/api/v2/auth/refresh session_id=sess_6Kp3 refresh_jti=rt_4b21 retry=1 2023-10-02T16:07:11.644Z level=ERROR service=auth env=staging request_id=req_stg_7f3c2 op=refresh_exchange result=reject code=session_version_mismatch expected_session_version=27 got_session_version=26 expected_refresh_jti=rt_4b22 got_refresh_jti=rt_4b21 2023-10-02T16:07:11.645Z level=ERROR service=api-v2 env=staging request_id=req_stg_7f3c1 op=retry_original_request result=abort status=401 error=refresh_retry_failed_after_successful_clerk_revalidation 2023-10-02T16:07:11.646Z level=WARN service=web env=staging request_id=req_stg_7f3c1 op=response_finalize set_cookie_session_version=27 set_cookie_refresh_jti=rt_4b22 note=response_finishes_after_failed_retry 2023-10-02T16:07:12.991Z level=INFO service=web env=staging request_id=req_stg_7f3c4 method=GET path=/api/v2/workspaces/ws_19/sources user_id=user_447 session_id=sess_6Kp3 auth_jwt_status=fresh auth_jwt_jti=jwt_new_92 source=hard_reload 2023-10-02T16:07:13.041Z level=INFO service=api-v2 env=staging request_id=req_stg_7f3c4 result=200 My read is we're rotating in Clerk, but the retry path is still pulling stale auth context from cache before the new cookie/session version lands. Only saw it on requests that expire mid-flight. Clean login and clean refresh still pass. Read the logs and DM Marcus in our internal team chat with a short diagnosis and the questions he needs to answer before he calls the patch green. My read is stale cached auth context after Clerk revalidation but before the new cookie/session version lands; sanity-check that and keep the note focused on the retry failure mode, not a broader auth redesign.

Devon’s first cut for Elena is here: From: Devon Hayes <devon@atlas-test.com> To: Morgan Chen <morgan@atlas-test.com> Subject: first cut / Elena Mercury calibration Date: Tue, Oct 3, 2023 7:42 AM PT First cut below. This is meant as a calibration note, not a process-launch deck. Main places I'd like your pass: - whether the pricing section is too detailed for a first pre-read - how directly to say expansion is uneven without making the whole thing feel like a shrug - whether we say Evergreen by name in the pre-read or keep that for the conversation --- Mercury / October calibration draft for Elena Volkov Why this note We are considering a selective October reopen, starting with a short list and using this as a calibration conversation rather than a broad process update. The goal is to test whether the Mercury story is now honest and credible enough to reopen narrowly. What has changed since February - Product signal is less theoretical. The design-partner wave produced real evidence on setup, activation, and admin friction. - The activation/admin story is better than it was earlier in the year because the team corrected the biggest early setup and handoff failures. - Pricing is now structured instead of bespoke. - Evergreen put us through a real regulated-buyer procurement/security process, which surfaced enterprise-readiness gaps with much more specificity than we had before. September Mercury package — usable summary Board-facing context - The quarterly cohort view is directionally better in the recent design-partner wave than in the earliest Mercury cohorts. - That improvement is useful context, but it is not proof that retention or expansion is solved. Operating read - Activation improved after the onboarding/admin fixes in the design-partner wave. - Admin friction came down in the latest wave. - Expansion remains uneven: some accounts deepen after setup, others stall after initial usage. - The package is best used as evidence that the team improved the early path and clarified the failure points, not as proof of a finished growth engine. How to talk about Evergreen What Evergreen gives us - a real buyer process in a regulated environment - concrete evidence on where Mercury still needs deeper enterprise-readiness work - a better read on packaging gaps around security/admin expectations and procurement handling What Evergreen does not give us - proof that enterprise readiness is solved - proof that the motion is repeatable across multiple accounts - proof that procurement converts cleanly into expansion Pricing frame Pilot - $2,500/month platform minimum - includes up to 50 monthly active developers Growth - $7,500/month platform minimum - includes up to 200 monthly active developers Overage - $1,000 per additional 50 monthly active developers Enterprise add-ons - SSO, audit logs, and advanced admin controls are separate add-ons only after those features ship - we should be explicit that those are not being pre-sold as current capability Caveats to keep explicit - Expansion is still uneven. - Evergreen is a strong signal, but still a single signal. - Enterprise-readiness gaps are clearer than before; they are not closed. - This would be a narrow October reopen, not a broad fundraising process. - I do not think we should put random investor coffees on the calendar off this package. Questions to test with Elena 1. Is this package materially more credible than the February version? 2. What would make the Evergreen signal feel repeatable rather than anecdotal? 3. Is the current pricing frame disciplined enough, or does it still read as early? 4. How much more explanation do we need on uneven expansion before widening beyond a short list? Rough email shell Elena — We took another pass at the Mercury package and wanted to use a short conversation for calibration, not to launch a broad process. The current story is stronger on three dimensions: the design-partner wave improved activation/admin friction, pricing is now more structured, and Evergreen gave us a real regulated-buyer signal on what enterprise readiness still lacks. The caveat is that expansion remains uneven, so we are not trying to hand-wave that away. If useful, I can send a tighter one-pager ahead of time and use the conversation to pressure-test whether this is credible enough for a narrow next step. Please turn this into a concise calibration pre-read and email it to Elena Volkov with my normal outbound signoff. This is not a process-launch deck. Use the September Mercury metrics, the hybrid pricing model, and Evergreen’s procurement/enterprise-readiness signal, but keep the caveats explicit: expansion is uneven, Evergreen procurement is not enterprise-readiness proof, and random investor coffees stay off the calendar. Separately, send Devon a prep note in our internal team chat with the same guardrails and the questions we want Elena to pressure-test.

The month-open AWS cost email shows non-prod snapshots running about $1,860 above the September run rate. Please DM Marcus in Discord and ask him to check the cause and send back one recommended cleanup for the snapshot overage. I only need one cleanup recommendation for this overage. Treat it as snapshot cleanup, not a new infrastructure project.

Priya’s Figma comments are the input: Figma comments — onboarding flow / Post-invite empty state v7 Frame: Empty state after invite acceptance, before first repo connect Date: Oct 3, 2023 9:11 AM — Priya Need copy call on this state before I hand it to eng. This screen is what a newly invited user sees when they've joined the workspace but nothing is connected yet. Important nuance: some users land here before they're actually ready to connect a repo, so the copy needs to be clear without sounding like we are upselling them inside the product. Current UI constraints: - one short headline - one support line - primary button stays 'Connect repository' - secondary text link stays 'I'll do this later' Option A Headline: Connect a repo when you're ready Support line: You can finish setup now or come back later from workspace settings. Option B Headline: Activate your team trial Support line: Connect a repo to unlock the rest of the setup and invite your team. My read: - A is calmer and more accurate to the actual moment - B is punchier, but may feel salesy once the user has already accepted an invite and thinks they're already 'in' 9:18 AM — Jake From the Mercury side we still get the 'what am I supposed to do next?' question a lot on this step. A seems clearer if the user isn't ready immediately. B probably nudges action harder, but it reads a little marketing-y to me too. 9:24 AM — Priya Agree. Also worth noting we already removed the celebratory version because it made the empty state feel like a bait-and-switch when there wasn't data yet. I can ship either, but I'd like a decision today so I don't hold the next build. Give me a Figma-ready reply I can paste. Decision is Option A: “Connect a repo when you’re ready.” Reject “Activate your team trial” as too salesy for this moment. The rationale should be that the user has already accepted an invite, may not be ready to connect a repo yet, and we want a clear next step without making the empty state feel like marketing.

Marcus says the JWT retry patch is green and is asking how broadly to announce it. Please post a concise release note to #eng-releases only, not #eng-all: the API v2 auth refresh retry fix is live and no customer action is needed. Also DM Marcus in our internal team chat telling him not to post it in the broad engineering channel.

Maya texted that Mom is asking about Saturday plans. Please send Maya an SMS: Saturday is protected, Sunday brunch in Oakland at 11 works if Mom is flexible, and loop Jamie in only after Maya confirms.

Rishi’s Pinecone logs are here: From: Rishi Patel <rishi@atlas-test.com> To: Morgan Chen <morgan@atlas-test.com> Subject: Pinecone connector logs — header casing again Date: Wed, 4 Oct 2023 09:42:18 -0700 Morgan — Forwarding the Pinecone excerpt below. The odd part is that preserved `X-Scaffold-Signature` and fully lowercased `x-scaffold-signature` both pass, but support is still seeing people assume any normalized variant is fine. Can you sanity-check whether the docs need one tighter line? I do not want to accidentally imply we accept underscore/env-var style names or anything beyond the actual header contract. Forwarded log excerpt: ``` sandbox=atlas-connector-rerun-2 body_sha256=8eeb1f7c4d0b2d9c4cf0c6f7c4e0a9c1d8c6fd0b5f0a4c1e9a2d6b7c8f1e0a3 2023-10-04T15:58:12Z req=pc_84fa9 POST /v1/connectors/atlas/webhook headers: X-Scaffold-Workspace=atlas-sandbox; X-Scaffold-Signature=sha256=6f2c0f9d... verify: ok status: 200 2023-10-04T15:58:49Z req=pc_84faa POST /v1/connectors/atlas/webhook headers: X-Scaffold-Workspace=atlas-sandbox; x-scaffold-signature=sha256=6f2c0f9d... verify: ok status: 200 2023-10-04T15:59:11Z req=pc_84fab POST /v1/connectors/atlas/webhook headers: X-Scaffold-Workspace=atlas-sandbox; x_scaffold_signature=sha256=6f2c0f9d... verify: fail status: 401 error: missing required signature header `x-scaffold-signature` 2023-10-04T15:59:34Z req=pc_84fac POST /v1/connectors/atlas/webhook headers: X-Scaffold-Workspace=atlas-sandbox; X_SCAFFOLD_SIGNATURE=sha256=6f2c0f9d... verify: fail status: 401 error: missing required signature header `x-scaffold-signature` ``` Operator note from their side: same request body/workspace in all four attempts; header-only verification path on every run. Inspect this and draft a narrow response for Rishi. The fix should clarify that preserved `X-Scaffold-Signature` and fully lowercased `x-scaffold-signature` are accepted against the same body/workspace, but underscore/env-var style names are not. Do not introduce alternate signature locations or broader header normalization.

Just got out of the Elena calibration with Devon. We used the September Mercury metrics, the hybrid pricing model, and Evergreen's enterprise-readiness/procurement signal, with the caveats explicit: expansion is still uneven, Evergreen procurement is not proof that enterprise readiness is solved, and random investor coffees stay off the calendar. Elena said the story is materially more credible than February, but pushed on whether Evergreen can become repeatable rather than anecdotal. Her warning was not to widen beyond a short list until the package can explain uneven expansion without hand-waving. Please reply to Elena in the existing email thread with a warm thank-you, copying Sarah, and use my usual in-thread signoff. Send Devon a short internal Discord summary with next actions. Also give me one board-note paragraph saying the B-round remains October prep with Elena as warm context, not a broad launched process.