Monthly run-rate estimate — 2026-06-04

Operator-requested multi-month vendor sweep + run-rate estimate Scope: 2026-03-06 through 2026-06-04 (90-day window) Prepared: 2026-06-04 UTC

Blocker note — Gmail / Drive MCP connectors not available this session

The Gmail (mcp__claude_ai_Gmail__*) and Google Drive (mcp__claude_ai_Google_Drive__*) connectors were not available in this runtime. This means:

• Gmail could not be searched for invoices / receipts in the 90-day window.
• The Drive workbook (MooseQuest Ledger 2026) could not be read or written.
• No rows were added to or corrected in the Drive ledger this session.

Hard blockers for a full reconciliation:

1. Anthropic API usage and any Claude.ai subscription amounts — must be pulled from console.anthropic.com > Billing > March / April / May 2026.
2. Heroku May 2026 invoice total — confirmed as $216.62 from the prior session reference (2026-05-reconciliation-delta-heroku.md per operator briefing); the breakdown across apps is unknown without the PDF.
3. Postmark — approved out of sandbox 2026-05-09; billing tier and first charge date unknown without Gmail access.
4. Sentry — free-tier status vs. paid confirmed unknown without Gmail access.
5. Cloudflare — all products listed as free-tier currently; no charge expected, but unconfirmable without Gmail access.
6. Schwartz IP — engagement accepted 2026-06-02; invoice not yet billed per operator note in the task brief.
7. USPTO trademark filings — per docs/business/trademark-moosequest.md, fresh filings have not yet been made; fees are anticipated, not yet charged.

This document is therefore a best-effort estimate built entirely from operator-supplied context in the repository. Every line below is marked with its confidence level. The CPA should treat all figures as estimates pending Gmail / Drive reconciliation.

TL;DR monthly burn

Estimated recurring monthly run-rate as of 2026-06-04: ~$366–$386/mo

This is the floor. It does not include Anthropic API metered usage (variable), Postmark (billed by email volume, unknown), any Schwartz IP retainer invoices not yet received, or any USPTO filing fees not yet charged.

Run-rate table

Heroku breakdown estimate (May 2026 invoice $216.62)

The May invoice total of $216.62 is operator-confirmed. The cost analysis in docs/business/2026-05-04-velvet-908-cost-analysis.md gives the following baseline layout:

Reconciliation note: raxx-api-prod was upgraded from Essential-0 to Standard-0 on 2026-05-15 (PR #2171). The May invoice would reflect a prorated Standard-0 charge for Raptor prod. This $50/mo line is new starting mid-May; full impact shows in June.

CPA flag: Without the itemized Heroku invoice PDF, the per-app Postgres and dyno breakdown cannot be confirmed. The operator should pull the May PDF from dashboard.heroku.com > Billing > May 2026 and upload to MooseQuest > Finance > Expenses > 2026 > 2026-05/.

Methodology

How recurring costs were derived

• Google Workspace ($22/mo): Confirmed from 2026-04-reconciliation.md — billing dashboard read 2026-04-22; Business Plus 1 seat $22/mo on annual commit.
• Heroku (~$216.62/mo): May invoice total confirmed by operator in task brief. Used as the single-month baseline for the recurring estimate since velvet infra was provisioned mid-April/early-May and the June bill will reflect the full load.
• Dreamhost ($5.00/mo): Confirmed monthly billing from 2026-04-reconciliation.md.
• GitHub ($4.00/mo): Pro plan confirmed in project_ci_billing memory note.
• Oracle Dyn (~$5/mo): Estimated from prior-period pattern; not confirmed by invoice.
• Apple Developer ($8.25/mo annualized): $99 annual fee paid 2026-04-22; annualized for run-rate purposes as $8.25/mo.

How one-times were counted

• Only expenses with confirmed invoices or confirmed order records are listed in YTD totals.
• Anticipated expenses (Schwartz IP, USPTO) are called out as anticipated but NOT included in the run-rate or annual projection totals.
• The Cloudflare domain registrations ($95.58 total, April 2026) are one-time annual registrations; they reappear as renewal costs in 2027.

Annualization

• Annual projection = monthly recurring × 12 + confirmed one-time YTD (not double-counted against monthly).
• The Apple Developer Program annual fee is counted once in the annual total, not × 12.

Monthly + annual projection summary

Confirmed recurring monthly (floor, excludes unknowns):

Note: Apple Developer amortized $8.25/mo is included in the annual projection only.

Confirmed one-time / annual YTD:

Annual projection:

Sensitivity: what changes when the first customer signs up

1. Stripe fees appear. Not a cost until revenue exists. Per vendor-conventions, categorized as Commissions & Fees. Rate: 2.9% + $0.30 per transaction. At $29/mo Founders plan, each card charge nets ~$28.86 after Stripe. Budget as variable Commissions & Fees; not in current run-rate.

2. Heroku may scale. The current dyno formation (Eco/Standard-1X across several apps) can absorb low-single-digit customer traffic. The Raptor Standard-0 Postgres handles up to 200 connections. No forced scaling event at early customer counts — but worth monitoring at ~50+ concurrent active users.

3. Sentry may cross the free-tier threshold. At 50k errors/mo the free tier holds. If Raxx onboards multiple active users who generate errors, monitor the error event count. Team tier ($26/mo) is the next step.

4. Postmark billing activates. Pre-launch, Postmark volume is near zero (ops alerts only). Once transactional email fires for real customers (welcome, 2FA, billing receipts), the volume climbs. The Postmark free tier is 100 emails/mo; the first paid tier is ~$15/mo (10k emails/mo). Budget $15/mo as a soft ceiling for early-customer volumes.

5. Anthropic API usage grows. If the Confidence Engine AI features see customer use, the API meter increases. Currently the only usage is agent-dispatched work (operator-facing). No customer-triggered API calls in current architecture — confirm before launch.

6. Northwest registered-agent annual renewal kicks in 2027-05-13 at $125/yr. Not a 2026 cost; note in calendar.

7. PA Annual Report ~$7 in 2027. Minimal; file via PENN File. Not a 2026 cost.

Candidate cuts (observations only — no action without operator review)

1. raxx-velvet-staging Standard-0 Postgres. The 2026-05-04 cost analysis (docs/business/2026-05-04-velvet-908-cost-analysis.md, Section 3b) explicitly recommended downgrading the Velvet staging Postgres from Standard-0 ($50/mo) to Essential-2 ($20/mo), saving $30/mo ($360/yr). If this downgrade was not applied when Velvet staging was provisioned, it is the highest-confidence savings candidate. Confirm current tier via heroku pg:info --app raxx-velvet-staging.

2. raxx-api-staging Standard-0 Postgres. Raptor staging does not need point-in-time recovery. If staging Raptor is also on Standard-0 (the prod tier), downgrading to Essential-2 saves $30/mo. Confirm tier via heroku pg:info --app raxx-api-staging.

3. Idle Standard-1X dynos on raxx-velvet-staging. Velvet staging is operator-triggered only (rotation jobs are not continuous). Downgrading staging dyno from Standard-1X ($25/mo) to Eco ($5/mo) saves $20/mo, with the tradeoff of cold-start latency (5–30 sec) when staging rotations are triggered. This was flagged in the cost analysis as architecturally incompatible for prod (the dyno must be always-on for prod) but may be acceptable for a staging-only rotation rehearsal path where latency is not customer-visible.

4. Dreamhost shared hosting. At $5/mo, this is low-priority to cut. However, if the Dreamhost hosting is serving only static pages that could be served by Cloudflare Pages (free tier), it may be eliminatable. Flag as low-priority observation — confirm with operator what Dreamhost currently serves.

5. Oracle Dyn DNS (~$5/mo). Per feedback_dyndns_stays.md, Dyn also serves the operator's home network — migration is off the table. Listed here only for completeness; do NOT propose migration. This is a confirmed stay-put decision.

CPA flags

The following items require CPA judgment at year-end. Listed in priority order.

1. Pre-LLC expenses (Jan–May 2026). All expenses before MooseQuest LLC formation (2026-05-13 order date; ~2026-05-25 to 2026-05-30 expected PA approval) are sole-proprietor Kristerpher Henderson Schedule C expenses, Entity = sole-prop-kris. The CPA should confirm: (a) which date triggers the Entity shift to MooseQuest LLC; (b) whether any pre-formation operating expenses can be captured as start-up costs under IRC §195 instead; and (c) how Heroku/cloud-infra expenses during the development-only period (pre-revenue) are classified.

2. FastSpring / FreeScout $105.75 lifetime licenses. Three open questions documented in docs/business/expenses/2026-05-03_FastSpring_FreeScout_modules.md: (a) de minimis safe harbor — does each module qualify individually or are they aggregated; (b) PA use-tax exposure on FastSpring's collected sales tax rate; and (c) pre-formation expense treatment (Schedule C vs. IRC §195 start-up cost). Deductible column = confirm-with-CPA until resolved.

3. Northwest LLC formation — expense allocation. The $214 Northwest order contains three components with different tax treatment: $39 service fee (likely Legal & Professional Services), $125 PA state filing fee (Taxes & Licenses), and $50 EIN service (likely Legal & Professional Services or Taxes & Licenses). The CPA should confirm the split and whether any portion is a capitalizable organizational cost under IRC §248. Recommend three separate ledger rows with the same source.

4. Heroku vendor risk + depreciation posture. The cost analysis doc surfaces Heroku's sustaining-engineering status (Salesforce enterprise sales ended Feb 2026). If a migration to AWS occurs in 2027, the CPA should advise on whether migration costs are expensed in the period or capitalized. Not a 2026 action; noted for planning.

5. Apple Developer Program $99 — cash-basis year recognition. Paid 2026-04-22 for a 12-month enrollment. On cash basis, fully deductible in 2026. The CPA should confirm this is correct and that no accrual treatment is warranted. Category = Dues & Subscriptions per vendor-conventions.

6. Cloudflare getraxx.com 3-year prepaid registration $31.38 (paid 2026-04-22). Covers 2026-04-22 through 2029-04-21. On cash basis, the full $31.38 is deductible in

7. If/when the entity switches to accrual post-LLC, the CPA should advise on amortization ($10.46/yr over 3 years). Noted in the 2026-04-reconciliation.md.

8. Anthropic API — project allocation. When Anthropic charges are pulled, each month's charge should be allocated between MooseQuest consulting (operator-agent calls serving consulting work) and Raxx product (Raptor/Antlers/Console agent calls). The split is unknown without console.anthropic.com usage data. Likely split: majority Raxx product during this development period.

9. Entity boundary date. The ledger transition from sole-prop-kris to MooseQuest LLC needs a confirmed date (PA DOS approval certificate date, not the order date). The bookkeeper will update the entity column on all rows after that date once the certificate arrives.

Open items for operator (action required)

1. Pull Anthropic billing for Mar / Apr / May 2026. Log into console.anthropic.com > Billing. Capture the monthly total and any Claude.ai Pro/Max subscription charges as separate line items. Paste or forward to the bookkeeper.

2. Pull Heroku May 2026 itemized invoice. Log into dashboard.heroku.com > Billing > Invoices. Download the May PDF and upload to Drive at MooseQuest > Finance > Expenses > 2026 > 2026-05/. This unlocks the per-app breakdown. Also confirm: is raxx-velvet-staging on Essential-2 or Standard-0 Postgres?

3. Confirm Postmark billing tier. Log into account.postmark.com > Billing. Confirm whether the account is on the free tier (100 emails/mo) or a paid plan. If the sandbox approval on 2026-05-09 triggered an automatic paid-tier enrollment, there may be a charge to book.

4. Confirm PA DOS approval date for MooseQuest LLC. The certificate of organization approval date from Northwest/PA DOS is the trigger for the Entity column transition in the ledger. Once received, forward the certificate date.

5. Schwartz IP engagement — invoice expected. The IP assignment engagement was accepted 2026-06-02. When the invoice arrives, book as Legal & Professional Services, Entity = MooseQuest LLC, Project allocation = Shared / parent entity.

6. Confirm Oracle Dyn monthly rate. Invoice amount ($5/mo estimated) needs to be confirmed against an actual Dyn billing statement. Forward or paste the current plan rate.

Summary of changes to Drive ledger this session

None. Gmail and Drive MCP connectors were not available. The following rows need to be added to the Drive Expenses tab when access is restored:

All "NO RECEIPT — flag" rows must remain flagged until the PDF or dashboard screenshot is uploaded to Drive. Do not convert to confirmed without a receipt.

Vendor sweep — vendors not found / not expected

The following vendors from the requested sweep are not expected to have charges in the 90-day window and are documented here to close the loop:

Vendors not pre-listed that were surfaced

No surprise vendors were identified from repo docs alone. However, the following items warrant attention if Gmail access is restored:

1. Redis add-on on Heroku (raxx-api-prod). WebAuthn challenge store added in PR #2728 per docs/ops/runbooks/heroku.md. A REDIS_URL is referenced. Heroku Data for Redis mini tier is $3/mo; the standard tier is $15/mo. This may be a new line in the Heroku invoice not separately tracked. Confirm tier and whether it is rolled into the $216.62 May total or is a separate line.

2. Heroku log drain / alerting. The 2026-05-11 incident doc references a Heroku log drain. Some third-party log drains have their own cost. Confirm whether any external log-drain SaaS (Papertrail, Logtail, etc.) is attached to any Heroku app. If so, add to ledger.

3. Domain registrar for moosequest.com. Per docs/business-legal/domain-moosequest-com-registrar-research-2026-05-13.md, the moosequest.com domain was under investigation. If it was purchased, it would be an Internet & Hosting or Advertising expense depending on intended use. Cannot confirm without Gmail access.