PropFlow: cash & cards — the plan
CFO summary for the team · 2026-08-02 · every number sourced or labeled an estimate; the full research is in the appendix at the bottom
Recommendation
- Invest the idle cash in U.S. Treasuries: keep $100K in checking (~3 months of burn), move $320K into VUSXX — Vanguard's Treasury money-market fund — earning ≈ $980/month. Same/next-day liquidity, backed by the U.S. Treasury, and it ends our ~$170K of uninsured single-bank exposure.
- Card: Mercury IO — 1.5% back on everything, $0 cost, no personal guarantee. Team vote below (Part 2).
- Claim ≈ $19K of startup credits — the GitHub credit alone takes our single biggest software line (Actions, ~$620–1,000/mo) to ≈ $0 for about a year.
Current state → proposed state
Today
- $420K sits in one checking account earning $0 APY assumed 0 — confirm
- ~$170K of it is uninsured (over the $250K FDIC limit at one bank, if no sweep)
- Software runs $3–5K/mo, all paid at full price from checking — no rewards, no credits
- Income from assets: $0/yr
Proposed
- $100K checking + $320K VUSXX → interest ≈ $980/mo ($11.8K/yr)
- Checking under the FDIC limit; the invested $320K's credit exposure is the U.S. Treasury itself
- ≈ $19K of credits offset the software bill (GitHub, AWS, ElevenLabs, Anthropic)
- 1.5% back on every card dollar (+$45–75/mo at current spend)
Monthly run-rate — before and after
| Line | Today | With the plan |
|---|---|---|
| Payroll (3 people) | ≈$30,000 | unchanged |
| GitHub (Actions + seats) | $650–1,060 | ≈$0 for ~12 mo (credits) |
| Anthropic (9 subs + API) | ≈$1,900+ | reduced (startup credits) |
| AWS + ElevenLabs | in the $3–5K pool | ≈$0 while $9K credits last |
| Other SaaS (Slack, Twilio, …) | in the $3–5K pool | −1.5% cashback |
| Cash income | $0 | +$980 |
Stated run-rate + spot-verified charges; sourcing and line-by-line detail in the appendix.
Part 1 — Treasury: which one, and why
VUSXX (Vanguard Treasury Money Market Fund), held in a Vanguard organization brokerage account. 7-day yield 3.68% (7/31), expense ratio 0.07%, $3K minimum, opens online for a Delaware C-corp, same/next-day liquidity. Government-only and short-duration — this is cash preservation with yield, not an investment position. Keep ~3 months of burn (~$100K) in checking (Kruze's standard for startups is 2–3 months) and top it back up monthly.
On $320K that's $11.8K vs $10.8K vs $9.6K per year — Vanguard wins because it charges 0.07% while the fintech platforms take ~0.6% or pay via lower-yield funds. Full 9-product comparison, the headline-rate traps (4%+ quotes that aren't cash-equivalent), and split sensitivity: appendix.
Part 2 — Cards: the decision, by the math
| Mercury IO | Meow | Brex | |
|---|---|---|---|
| Annual fee / cost per employee | $0 / $0 | $0 / $0 | $0 / $0 |
| Earn rate on our stack | 1.5% on everything | 2.5% Anthropic+OpenAI · 1.5% else · 0.5% AWS | 1.2–1.8% SaaS · ~0.6% cloud/AI |
| Cashback / yr est | $630 | $786 | ~$320–460 |
| Perks that apply to us | ≈$19K one-time | none found | ~$650–1,450/yr + $500 est |
| Year-1 total value | ≈$19.6K | ≈$0.8K | ≈$1.5–2.4K |
| Strings | limit = Mercury balances | must bank at Meow; rates changeable | weakest earn; AI category unresolved |
With 9 Anthropic subs, Meow now edges pure cashback (+~$156/yr) — the case for option B. Mercury still wins year-1 by ~$19K: the credits attach to its bank account, free, with no card spend required. Split assumptions + full card detail: appendix.
Proposed — needs team decision. Pick one (your choice is saved under your name; everyone sees the votes). Fede's stated lean: one card only (8/2). The evidence for each option is in the sections below.
Treasury plan (Part 1):
The perks — every one that applies to us
| Perk | Value | Claim via | How |
|---|---|---|---|
| GitHub for Startups | $10,000 | Mercury (partner-only) | Covers seats + Actions minutes; zeroes our biggest software line ~12 mo |
| ElevenLabs Scale grant | ~$4,000 | Mercury | 12 months + 33M chars |
| AWSAWS Activate | $5,000 | Either platform | Once ever — check Activate console first |
| Claude for Startups | unpublished | Direct only | claude.com/programs/startups; first-party API only |
| #Slack discount | 30% / 25% | Brex / Mercury | Applies at renewal |
| Gusto 50% off yr-1 | ~$650–1,450/yr est, combined | Brex only | Gusto's software fee, not wages |
| QuickBooks 30% × 12 mo | (in est above) | Brex only | If we adopt QuickBooks |
| v0 credits | $500 | Brex only | Usable on our Vercel stack |
Totals: Mercury ≈ $19K one-time · Brex ≈ $650–1,450/yr est + $500. Every grant is once per company — claim the better version of each, once. Claim order, tripwires, fine print: appendix.
Execution checklist
- Open a Vanguard organization account (online; formation docs + EIN) → transfer $320K → buy VUSXX. Monthly calendar item: top checking to ~$100K after the second payroll.
- Open a free Mercury account → hold ~$15–30K float → issue 3 IO cards with per-card limits → move Anthropic, OpenAI, AWS, Vercel, Twilio, ElevenLabs, GitHub, Workspace, Slack billing onto them.
- Claim credits in the order above. Meow / Brex remain on hold pending the team vote.
- Confirm with the bank: actual product name, APY, any FDIC sweep. Write a one-page investment policy (govt MMF/T-bills only, who can move money).
- Revisit on: a Fed move, the next raise, burn materially above $33–35K/mo, or the VUSXX-vs-platform gap closing (suggested triggers).
Appendix — the full research (rates, all 9 treasury products, card details, rejected field, fine print, provenance)
The position — inputs
Inputs: $500K raised; current cash $420K stated by Fede 8/2 (the 7/30 bank-app screenshot showed $439,011.15 in one account — "Chase Business Complete" is inferred from the header, not confirmed); $30K/mo salaries (stated; corroborated by the 7/30 Gusto run of $14,487.72 — two such cycles ≈ $29K/mo); software $3–5K/mo, $3K low bar (stated). AI/SaaS charges (Anthropic $210.30, OpenAI, Slack) currently debit checking directly — earning nothing. GitHub Actions ran $618–998/mo in the July CI audit (+$63/mo Enterprise seats at list). Fede reports 9 Anthropic subscriptions (8/2) — ≈$1.9K/mo if all at the observed $210.30 tier (mix unconfirmed). A true 3-month ledger pull is the one missing artifact if the team wants exact splits.
Treasury — the full field
Net yield at PropFlow's tier (government-cash products, sourced + dated)
Why direct Vanguard beats the fintech treasuries: the gap is mostly fees — Mercury and Rho charge 0.6%/yr at this tier; Brex adds no fee but pays via a lower-yield fund plus a revocable Brex-paid top-up; Fidelity's funds carry a 0.42% ER vs VUSXX's 0.07%. Vanguard explicitly serves corporations ($3K fund minimum) vs minimums of $250K (Mercury), $100K (Rho, Meow). Two of six fintech vendors changed owners in the last four months (Brex → Capital One, closed 4/7/26; Arc → Axos, announced 7/7/26).
Headline-rate traps (not cash-equivalent): Rho's 4.10–4.55% column is a corporate bond fund; Ramp's "up to 4.33%" is a yield-to-maturity quoted on a $10M sample; Arc's "up to 4.59%" is gated behind paid tiers; Meow's 3.79–3.96% is commercial paper (credit risk) and needs $100K idle in Meow checking. Stay short and government-only: the Fed has held at 3.50–3.75% for five meetings and July's three dissents wanted a hike.
Split sensitivity: $70K/$350K ≈ $12.9K/yr · $100K/$320K ≈ $11.8K/yr (recommended) · $150K/$270K ≈ $9.9K/yr — differences are small; operational comfort wins.
Yield in context: $11.8K/yr is floating (declines as cash draws down and if the Fed cuts); interest is taxable at the 21% federal corporate rate → ≈ $9.3K net. The SVB lesson behind the insurance point: 94% uninsured single-bank deposits, $40B out in one day.
Cards — full detail
| Card | Rate on this stack | Cost | PG | Catch |
|---|---|---|---|---|
| Mercury IO — primary | 1.5% flat, everything | $0 | None | limit = Mercury balances |
| Meow — AI overlay | 2.5% Anthropic/OpenAI · 1.5% else · 0.5% AWS | $0 | None | must bank at Meow; daily-settled collateral |
| Brex — perks play | 2x SaaS (=1.2% at 0.6¢/pt) · cloud/AI likely 1x | $0 | None | weakest earn; perks are the point |
Mercury IO — the primary card
Flat 1.5% cashback on all spend, auto-deposited monthly, no exclusion list — coding-immune, so AWS and Anthropic earn full rate. $0 annual, cards "as many as needed" free, per-card limits/merchant locks/category restrictions, no PG, no personal credit pull. Charge card from Patriot Bank via Mercury (banking at Choice Financial + Column, sweep-network FDIC): limit tracks total Mercury balances (support docs describe counting Plaid-linked external accounts too — confirm at application); $15K in balances unlocks monthly terms. Card FX 3% (the "1%" seen in reviews is Mercury's wire fee). At $3–5K/mo → $540–900/yr. Perk pages advertise $5K AWS credits, $10K GitHub credits, a year of ElevenLabs, 25% off Slack (confirm live terms at claim). Do: open the free account, hold ~$15–30K working float (treasury stays at Vanguard — Mercury's own treasury nets 3.01%, skip it), issue three cards with per-card limits, move every software billing method over.
Meow — the AI-vendor overlay (conditional)
The only program our 20+-card sweep found with a contractual AI-vendor rate: legal reward terms name Anthropic and OpenAI at 2.5%, everything else 1.5%, AWS 0.5%, Walmart/Costco 0.01% (verified directly, 8/2). $0 fees, unlimited cards with per-card limits, no PG/credit check; issued by Cross River; limit proportional to your Meow checking balance, balance auto-paid daily; FX up to 3%; cashback must be claimed monthly in the dashboard. Rates changeable at Meow's discretion. The decision rule (rewards arithmetic): Meow beats a flat-1.5% card iff monthly Anthropic+OpenAI spend exceeds AWS spend (+1% on AI, −1% on AWS, tie elsewhere). At an illustrative $600 AI / $900 AWS / $1,500 SaaS split illustrative it yields ~$504/yr — losing to Mercury's $540. Update (8/2): Fede reports 9 Anthropic subscriptions ≈ $1.9K/mo — the condition now likely holds. Main-table cashback math at the informed split ($1.9K Anthropic / $600 AWS / $1K other SaaS ≈ $3.5K/mo, AWS+SaaS still illustrative): Mercury $630/yr · Meow $786/yr · Brex ~$320–460/yr. Do: audit one month of real bills; adopt only as the Anthropic/OpenAI billing card with a small checking float, or skip.
Brex — the perks play (optional)
Standard startup card, EIN-only, no PG, $0 on Essentials. Now a Capital One company (announced 1/22/26 at $5.15B; closed 4/7/26 — verified via Capital One newsroom + SEC filings; continuity messaging is stated intent, not contract). Earn is the weak spot: 2x "recurring software" (3x on the daily product per Brex's support grid) at 0.6¢/pt cash = 1.2–1.8%, while Brex's own category docs put AWS-class cloud in a separate "Servers" bucket absent from the bonus table → likely 1x (0.6%); AI-usage billing unresolved — ask in writing. Points transfer 1,500:1,000 to seven airlines (worth it only if you book award travel). Monthly product needs a $50K cash balance; daily requires banking at Brex (limit = balance). The reason to open it would be perks — but see the perks section: the deep-dive found Mercury holds equal-or-better versions of the big three (GitHub $10K, ElevenLabs 12-month grant, the same once-ever AWS $5K), leaving Brex's real exclusives for this stack at Gusto 50% yr-1, Slack 30% (vs 25%), QuickBooks 30%×12mo, and $500 of v0 credits; each Brex perk is once per account, and every major vendor grants once per company. No public signup bonus live as of 8/2; referral/accelerator offers exist — ask.
Also evaluated — the rest of the field
- Amex Business Gold ($375): dated community datapoints show 4x actually triggering on Anthropic, OpenAI, Twilio — but AWS earns 1x per multiple dated user reports incl. two written Amex-support replies (AWS is absent from Amex's published eligible examples), Google Workspace datapoints negative/mixed, points cash out at 0.6¢, employee cards cost extra, and it needs a personal guarantee + hard personal pull. Rarely clears its fee at this scale without award-travel redemptions. (Its $300 ChatGPT credit is real but covers ChatGPT Business only — a competitor product.)
- Ramp: cashback undisclosed — officially 0–1.5%, set per customer post-approval (flat 1.5% ended May 2024); "5%" marketing is modeled savings, not an earn rate. Unrankable.
- Slash (2% Pro, $25/mo): rate leader but Amazon + Microsoft are excluded merchants with AWS/GitHub inheritance unresolved; 0% on foreign transactions.
- Rho: up to 1.5% but must bank at Rho; miss one on-time payment → zero that period; rewards expire in 12 months.
- Expensify (1%), BILL Divvy (~1% effective + $5K bonus cap + 30%-utilization rule + 12-mo lockup), Stripe (Stripe-payments customers; underwriting tied to processing relationship): dominated.
- Not buyable / wrong product: Torpago (closed to direct applicants), Center (acquired by Amex), Airbase (inside Paylocity), Coast (fleet-gated), Keep (Canada), Rain (crypto-collateral), Arc (mid-acquisition), Every (0.5% at this balance tier). No issuer anywhere has an "AI" bonus category besides Meow's named merchants.
Free credits — fine print deep-dive 8/2
| Credit | Value | Best path | Gate to clear |
|---|---|---|---|
| GitHub for Startups | $10,000 | Mercury — application is partner-only (Mercury is a program partner; no solo route). Credits cover Enterprise seats + metered usage incl. Actions minutes ("eligible add-ons… and other metered offerings" — GitHub, 8/2), 12-month validity. Enterprise also includes 50K Actions min/mo (vs 3K on Team) — directly attacks our largest software line (Actions ran $618–998/mo in the July CI audit) | Outside funding ≤ Series B (GitHub's wording — angel SAFEs count); new to GitHub Enterprise (incl. past 6 mo); no prior GitHub credits; once ever |
| ElevenLabs Scale grant | ~$4,000 | Mercury (12 months + 33M chars, code on their perk page) — Brex's 3-month variant is strictly worse | <25 employees ✓; one per company; requires a "Grants" badge on your site 12 months |
| AWS Activate | $5,000 | Either platform — identical pool, claim exactly once | Paid-tier AWS ✓, pre-Series B ✓, no prior redemption (your 12-month history shows $0 applied — confirm in the Activate console). Strategy: re-applications only pay the delta — if an investor can unlock a $25–100K Activate Portfolio Org ID, take that instead and skip the $5K |
| Anthropic — Claude for Startups | unpublished | Direct only (claude.com/programs/startups) — on no card platform | Institutional equity, <4 yrs old, no prior Anthropic credits. Note: credits apply to the first-party Claude API only, never Bedrock/Vertex |
| Slack 30% / Gusto 50% yr-1 / QuickBooks 30%×12mo | ~$650–1,450/yr estimate | Brex — its genuine exclusives for this stack (+$500 v0 credits) | Requires opening a Brex account; each perk once per account |
| Datadog $100K · Auth0 12mo · Algolia $10K · OpenRouter $1K · Infobip $50K | $0 today | Mercury — only if the stack adopts these tools | Mostly new-customer-only; OpenRouter needs <$500 lifetime spend |
- No double-dipping, anywhere. Vendor-side terms gate every major grant at one per company (GitHub: "may redeem… only once"; ElevenLabs: "one application per company"; AWS: upgrade-only deltas; Anthropic: "not previously received"). Brex additionally caps each perk once per account. Pick the better version of each perk and claim it from that platform only.
- Two tripwires: accepting a small Google Cloud credit anywhere kills eligibility for the $200K GCP Scale tier later (its ceiling is $5K prior credits); downgrading GitHub before credits are consumed forfeits the remainder permanently.
- Twilio has no startup-credit program (their own words) and no Browserbase or Google Workspace credit exists on any platform surveyed.
- Implication for the card vote: this strengthens option A — Mercury alone captures the top ~$19K with zero card-spend strings. The once-promised "is a no-usage Brex signup worth it?" answer: only for the Gusto/Slack/QuickBooks discounts (~$650–1,450/yr, estimates) — real but small; defensible either way, so it's noted here rather than argued.
Provenance
Method: six parallel research agents (Brex, Ramp, Mercury + field scan, Amex, merchant-coding, treasury) + two peer-session sweeps, all under source-URL discipline; key claims re-read from primary pages in-session; then a 149-agent adversarial verify pass over 148 claims (107 HARD; 31 flagged claims corrected or relabeled before publication; 10 estimates labeled). Rates are snapshots as of their stated dates and float.
Primary sources include: meow.com legal reward terms + card agreement · mercury.com credit/blog/agreement/pricing/perks · brex.com support (multipliers, categories, redemption, requirements, limits, perks) + capitalone.com newsroom + SEC 8-K/10-Q · americanexpress.com rewards-info + MR value page · github.com/enterprise/startups + github.com/pricing (8/2) · Arctic Shift Reddit archive datapoints (dated, labeled community) · federalreserve.gov (rates, SVB review) · treasury.gov daily bill rates · Vanguard/Fidelity fund APIs · SEC 485BPOS (Fidelity ERs) · fdic.gov · mercury.com/rho.co/ramp.com/arc/meow treasury pages · a16z + Kruze cash-management guidance. PropFlow figures: 7/30 bank-app screenshot (one account) + Fede's stated raise/salaries/software figures (2026-08-02); worked-example vendor mixes are labeled illustrative, not measured billing. Vendor logos: Simple Icons (monogram badges where a mark isn't redistributable).