Skip to content

Advisory — Operations & Documentation Best Practices

Purpose: Surface where current operating posture creates avoidable risk or audit noise, grounded in specific patterns visible in the Brex + Santander + Xero data already in this repo. Written as a memo from reviewer to operator — Austin can choose what to act on.

Scope: Posture and documentation. Not tax positions (see tax/advisory-tax-benefit-areas.md) and not contracts that should exist (see legal/advisory-contract-gaps.md).

Written: 2026-04-20. Not legal or tax advice.

The 30-second summary

The books will be clean going forward once Xero is live and vendor rules fire. The risk is the year-plus of historical activity that was recorded with ambiguous memos, no supporting records, and occasional channel leakage (guest revenue into Austin's personal WF 6248, LP capital calls from Brex Primary, SBA payments out of Five Points). None of that is unrecoverable — but each gap is an hour of research per item if challenged later. This memo identifies the highest-leverage places to tighten documentation now so the 2025–2026 audit-defensibility baseline is solid before the 5471, PFIC, and intercompany-services work layers on top.

1. Stop letting bank memos drive accounting treatment

Pattern observed: Brex's auto-generated "OWNER CONTRIBUTION" memo on the 8 JPMorgan → Santander MXN wires was taken at face value by prior bookkeeping. The memo is mechanically wrong (these are intercompany settlements, not capital contributions) and the repo has already corrected the framing. But the pattern generalizes.

Other auto-memo risks visible in the data: - GAVON RENFROE on 55 Brex Primary inflows (~$4.44M) — classified everywhere as "owner contribution," which is correct, but the memo string is brittle; a future JPM-sourced inflow that is not Austin's personal funds (e.g., RENFROE S-corp reimbursement, a trust distribution) would auto-classify the same way. - Edgewood Partner "SETTLEMENT" — reads like a lawsuit; is actually a utility-bill rebill. - RNIE FINES AND CINDY PAYO — compressed memo whose actual purpose is unresolved. Earlier inference of "severance + tax penalty split" was wrong (Austin direct 2026-05-09: no separation payout was issued to Cindy; the "PAYO" component does not refer to a Cindy payout). Treat this memo as a cautionary example of relying on bank wire memos for accounting classification — needs Austin/JME clarification. - Brex "Treasury" vs. "Vault" vs. "Primary" — Brex labels are account-naming conventions, not GAAP-meaningful classifications.

Best practice: 1. Every Xero transaction carries a purpose description added by the bookkeeper, not the bank memo. Format: [Entity] — [Purpose] — [Supporting doc reference]. Example: RHG — Intercompany settlement, Sep-Oct 2025 collections — WIRE-2025-09-24. 2. Build auto-categorization rules on counterparty identity, not memo strings. Pair rules with guardrails (amount ranges; frequency patterns) so unusual activity flags rather than auto-posts. 3. Lock the rule: no transaction >$10K posts without bookkeeper review of a source document. The 8 Brex→Santander wires, the 55 Primary inflows, and the $1.155M GK sweep would all be caught by this trip-wire.

2. Close the counterparty-identification tickets before Xero goes live

Pattern: vendor-mappings.md lists ~10 counterparties still flagged needs-identification. The largest is CO DASH LLC — $42,294 across 4 payments to Casa Moksha Dep — the single biggest unknown inflow in the Brex history. If Casa Moksha's 2025 guest revenue is reconstructed for JME without resolving CO DASH, the revenue-by-channel analysis is incomplete.

Priority triage (by amount and ambiguity):

Counterparty Amount Why it matters
CO DASH LLC (inflow) $42,294 2025 revenue recognition blocker
Richard Rosenblum $24,360 Revenue recognition; recurring guest or channel?
Jose Pablo Gonzalez Deleze (NVIO) $20,000 MXN Only non-RFH inbound on Santander MXN in 15 months
Bryan Cady short-term loan $15,000 Note receivable; no written terms documented
Yvette Ryan legal $5,000 Engagement scope unknown
Jana $350K "FUND HOLDING CO" → GK $350,000 Austin contests the memo; suggests a labeling error; material for 5471 ownership analysis

Best practice: Austin blocks 30 minutes, walks the list, either identifies each from memory or emails the counterparty. Any not resolved within a week gets a formal "unknown" tag in Xero with a quarterly re-review date. Do not let unresolved items bleed into year-end close.

3. Capture the capex-vs-expense decision at the point of spend

Pattern: Seven 2025 Casa Moksha vendors are flagged needs-JME-confirm on capex vs. expense — Don Sebastian (palapa), Paola Maria (water system enclosure), Geodrilling (drain well), Eran Generator (generator), Frigotek (HVAC), Nelsy (security cameras), Desarrollo Anjona (advisory). JME has been booking; the question is whether they capitalized each. If they expensed the palapa or generator (each a $20K+ project), cost-seg and basis are materially understated.

Why this is a best-practices issue, not just a JME question: the decision belongs to the operator who signed the invoice and knows whether it was "repair the existing" vs. "build new / add functionality." JME can only classify by what they see in the CFDI, which is often just a line-item description.

Best practice: 1. At the point of quoting/approving any Casa Moksha invoice >$5K, Austin (or Miroslava, with a template) writes a one-line "scope and effect" note: did this extend the useful life of an existing asset? Add a new function? Just keep it running? That note rides with the invoice into JME's intake. 2. Same for Brex-paid Mexican vendors (Megamak, Edy Daniel, Jorge Eduardo Jag, Desarrollo Anjona) — Xero can carry the note in the transaction description field. 3. At month-end, bookkeeper reconciles the "scope and effect" notes against JME's classification; discrepancies go to the monthly review call. 4. Capex inventory is regenerated from the tagged invoices, not reconstructed from bank memos a year later (see accounting/casa-moksha-capex-inventory.md).

4. Personal-card / personal-account business expenses need a cadence

Pattern: Two systemic commingling issues show in the data:

  1. Wells Fargo 6248 (Austin's personal checking) historically received Casa Moksha guest payments. Austin has stopped using it for this; historical statements are still needed to backfill 2024. accounting/personal-card-business-expenses.md likely expands on this.
  2. Reverse direction: five LP fund capital calls in 2025–2026 (ATP, Untapped, Crosslink, Garuda, Overwater) were paid from Brex Primary for personally-held investments. These were classified as business expenses at time of spend; the repo reclassifies them as owner distributions (confirmed 2026-04-20).

Exposure: without a regular reconciliation cadence, categorization errors compound. The LP payments were a month+ away from being silently expensed through Xero at setup.

Best practice: 1. Monthly personal-to-business reconciliation. On the 5th of each month, walk every Brex transaction that feels personal (LP payments, RENFROE-related items, personal travel, Austin's direct wire to Austin Marine for the slip transfer) and confirm its Xero classification. Fifteen minutes a month. 2. Corresponding business-to-personal: any Austin personal spend that was business-purposed (business travel on a personal card, meals with advisors) gets a single expense report into Xero monthly. Receipts attached. 3. One-direction rule: if a category gets crossed more than twice in 90 days (e.g., LP payments keep landing on Brex), raise as a routing fix — it's not a booking problem, it's an operations problem.

5. The $4.44M owner-contribution stack deserves a formal basis ledger

Pattern: 55 "GAVON RENFROE → Primary" inflows + $45K from WF 6248 + $100K same-day pair for Austin Marine + the $1.155M GK contribution + $350K from Jana → GK = a substantial and growing basis stack across at least three entities (RFH, Austin Marine, GK). Currently tracked implicitly through the Brex feed and Xero owner-capital account.

Risk: At liquidation or partial disposition, the basis question will be "what was contributed, when, and by whom?" The Brex feed is the system of record, but it's lossy — memo strings change, sub-account sweeps muddy the trail, and "GAVON RENFROE" doesn't distinguish between personal-JPMorgan, WF 6248, RENFROE reimbursement, or trust-sourced funds.

Best practice: 1. Maintain a standalone contribution ledger in /accounting/ as a CSV or markdown table: date, amount, contributor (Austin personal, Jana, RENFROE, trust), receiving entity, source account, purpose. Every capital movement adds a row. 2. Reconcile the ledger to Xero's owner-equity accounts at quarter-end. 3. Flag contributions from non-Austin sources (Jana's $350K, any future RENFROE S-corp payments) — these carry separate basis and potentially separate ownership-percentage implications, especially for GK where the 5471 will ask about ownership changes. 4. Re-confirm Jana's $350K status directly with Jana (Austin's mother); the Brex memo says FUND HOLDING CO which Austin contests. A five-minute call prevents a 5471 ownership mis-reporting.

6. Document the "one-time cash event" transactions now, not at audit

Pattern observed — large, rare, poorly-memoed transactions:

Transaction Amount Memo Supporting doc status
CHAMPIONSHIP TITLE → W 3101 +$1,201,892 "210 Lavaca Street # 3101/#3103" Closing statement needed (refi? partial sale?)
ServisFirst Bank ← Five Points -$2,325,725 "FCC: 1006 20th Stree" ServisFirst note purchase — documentation somewhere
MITSUI FUDOSAN ← Primary -$1,029,285 "Mita Garden Hills deposit" Mita closing statement needed
Renfroe Holdings GK → Primary +$1,155,644 (no memo) Mita financing flow — needs characterization
PNFP ← GK -$19,239 "May payment for short term line of credit" Bridge loan amortization needed
JANA P RENFROE JOINT → GK +$350,000 "FUND HOLDING CO" Contested memo
Warren Bartlett ← Austin Marine -$100,000 "Dock 8 Slip 7 ... Transfer Agreement" Agreement exists; attach to Xero record
Austin Renfroe Personal ← Primary -$815,936 "Repayment of RENFROE Shareholder loans" RENFROE-side corresponding record needed

Best practice: For every transaction >$100K, attach (or link to) the underlying document in Xero's file archive within 30 days of posting. Closing statements, wire instructions, loan agreements, transfer agreements. If a document doesn't exist, the 30-day window is when to create one. At $1M+ amounts, the "I'll find that later" pattern compounds fastest.

Simultaneously, draft a short transaction memo for each — two sentences explaining the business purpose and tax characterization — and attach it alongside the primary document. Future-you (or future-Pearce, or a future IRS examiner) is the audience.

7. Separate business and personal, even inside the family office ecosystem

Pattern — three kinds of commingling visible in the data:

  1. Personal ↔ RFH: ~$4.44M in (owner contributions), ~$844K out ($815K RENFROE loan repayment + $28K distributions + LP capital calls treated as distributions). Clean because RFH is disregarded, but the transaction volume means any misclassification looks like expense-vs-distribution and could be challenged.
  2. RFH ↔ sub-LLCs: capital contributions down, cash sweeps up (W 3101 $1.23M, others). Clean because stack is disregarded, but the W 3101 refi event ($1.2M title-company flow) was processed as a sub-account sweep without a capital-event memo.
  3. RFH ↔ RHG ↔ Austin personal (Mexico side): "OWNER CONTRIBUTION" wires with the wrong legal characterization. Fixed in the repo; needs to stop happening at Brex-memo-generation level (limited by Brex) and at Xero-categorization level (fixable).

Best practice: 1. Naming convention in Brex payee fields. Every manual wire should carry a payee memo that mirrors its tax characterization: "Intercompany settlement," "Capital contribution," "Owner distribution," "Loan principal," etc. Brex won't always preserve it perfectly, but the origin-side memo is the best defensive record. 2. Xero tracking-category discipline. Entity and Asset Class are already set up. Make sure every transaction posts with both dimensions populated. A transaction with an Entity tag and blank Asset Class is a reconciliation flag. 3. Quarterly intercompany balance confirmation. Every quarter, RFH's "Due to RHG" balance gets confirmed against RHG's "Due from RFH" on JME's books. Same with any RFH ↔ GK flows. A running 12-month reconciliation is the simplest defense against the Mexico or US taxing authority framing an intercompany as a deemed distribution.

8. The Miroslava classification question is an operations exposure, not just tax

Pattern: Miroslava receives a fixed $27,550 MXN/quincena (STP company portion + Mercado Pago personal portion). Fixed schedule + fixed amount + operational integration (property management at Casa Moksha) = Mexican labor law may deem this an employment relationship even if contracted as services.

Exposure if reclassified by SAT/STPS: - Retroactive IMSS / INFONAVIT contributions + penalties - Vacation pay, aguinaldo, profit-sharing (PTU) - Severance (3 months + 20 days per year served) if future separation isn't handled correctly - Cascading effect: if Miroslava is an employee, sub-contractors she supervises may be subject to outsourcing rules (Mexico's 2021 labor reform)

Best practice: 1. Ask JME (and ideally a Mexican labor lawyer — Portilla, Ruy-Díaz & Aguilar if their scope covers labor) for a classification memo. If contractor is correct, the memo becomes the defense document. 2. If employee is the safer characterization, convert proactively. Mexican labor authorities punish reclassification retroactively far more than voluntary conversion going forward. 3. Consider the same review for Alfredo, Rosi, Lucila — all are paid in biweekly quincena form, which is employment-like. 4. For the departed: Nano (through Jul 2025) and Cindy Canto — confirm separation documents are on file. Cindy especially given the back-tax catch-up history (she's the one person with visibility into books that were out of compliance during her tenure — clean-papered departure reduces future dispute surface).

9. The Cindy Canto handoff deserves a written record

Pattern: Austin confirmed (open-questions.md Q21) that separation handling needs to be verified. The 2025 JME back-tax catch-up work (two $15,022 MXN payments + the Sept 3, 2025 batch of 25 SAT federal tax debits) is the operational evidence of the bookkeeping gaps during her tenure.

Best practice — specifically for this situation: 1. Austin writes a short memo to file summarizing the circumstances of her departure: date, reason, what was found upon handover, what JME did to remediate. Dated and saved in legal/ or context/people.md. 2. Any signed separation documents get uploaded to the repo (or referenced if retained by counsel). 3. If none exist: Austin raises with Portilla, Ruy-Díaz & Aguilar — Mexican labor law gives former employees 1 year (generally) to bring labor claims. Austin should be past the window if she departed pre-2025, but confirming is cheap. 4. The JME back-tax catch-up itself is audit-relevant. Keep JME's work papers for the catch-up filings — they're the defense if SAT audits any of those periods.

10. Xero tracking is only as good as the inputs

Pattern: The single Xero org relies heavily on Entity + Asset Class tracking categories. Vendor mappings assume tracking dimensions are reliably populated. But the Brex auto-feed doesn't populate tracking — it lands in Xero with classification-only, and the bookkeeper (or auto-rule) adds the dimensions.

Risk areas: - Transactions that span entities: e.g., Consultoría y Servicios TI (used by both Casa Moksha and Oceana 433 per vendor-mappings.md). Invoice-level allocation is required; auto-rules can't do this. - Renfroe Innovation — no bank account; all activity is on RFH's Brex but tagged via the Entity dimension. If the Entity tag is dropped, the transaction silently posts as RFH. - Professional fees often span: Portilla, Ruy-Díaz & Aguilar is paid from both RFH and RHG depending on the invoice. Withers international tax work could be RFH or GK.

Best practice: 1. Pre-check rule: before Xero's monthly close runs, an automated query flags any transaction >$1K missing Entity or Asset Class tags. Bookkeeper resolves before close. 2. Invoice attachment discipline: for any professional-services or mixed-use vendor, the invoice PDF is attached to the Xero transaction. Avoids the "which entity was this for" question at reconciliation. 3. Renfroe Innovation watch: since it relies entirely on tagging rather than a separate bank account, run a monthly Renfroe Innovation report — lists every transaction tagged to it. Austin eyeballs: "yes those were all Innovation."

11. The "Tier 3 reporting" ambition needs an input discipline

Pattern: The new todo/tech-automation.md Tier 3 section (added 2026-04-20 from Claude Desktop) scopes a family-office report suite — financial position, income/cash flow, forecasts, investments, real estate ops, tax/compliance, governance. All downstream of Xero + Monarch + Brex + source data.

Best-practices observation: reporting quality is a function of categorization quality, and categorization quality is a function of input discipline. The reporting ambition is realistic if and only if items 1–10 above are handled. Otherwise the dashboards inherit the documentation gaps and present them more prominently.

Sequencing recommendation: 1. Close items 1–4 above (memo discipline, counterparty identification, capex tagging, personal-business reconciliation) — these are pre-requisite to trustworthy Xero reports. 2. Stand up Tier 1 Xero with auto-rules firing reliably for 2–3 months. 3. Then layer reporting on top. Starting reporting before the books settle yields charts that are mostly noise.

12. A short list of things the data suggests are going well

To avoid a purely corrective tone — a few things the operation already does well, worth protecting:

  • Entity hygiene at the sub-LLC level. Each sub-LLC has a dedicated Brex sub-account. Vendor-to-entity mapping works because the routing was set up correctly.
  • Separate Dep and Ops for Casa Moksha + Oceana 433. This is the single cleanest way to segregate guest deposits (liability until earned) from operating cash.
  • Use of Treasury + Vault for cash management. Interest income is meaningful (~$500+ per quarter based on Brex Treasury patterns) and it's on a dedicated account so the accounting is obvious.
  • The 2025 JME back-tax catch-up was done. That work is behind you, not ahead of you.
  • The collection-agent framing for Casa Moksha revenue. Legally cleaner than the alternative (RFH recognizes USD revenue, remits net to RHG) — the correct frame given RHG's Mexican tax posture.
  • The single-Xero-org decision. Adopting two orgs with consolidation journals would have tripled the maintenance surface for marginal value.

Action list (if this memo gets acted on)

Ranked by leverage × effort:

  1. Close the counterparty-identification tickets — 30 min of Austin's time, unblocks 2025 revenue reconciliation
  2. Jana $350K confirmation call — 5 min, prevents 5471 ownership mis-reporting
  3. Capex-vs-expense "scope and effect" note template — one-pager, changes operator behavior going forward
  4. Monthly personal-to-business reconciliation cadence — 15 min/month, catches commingling within 30 days
  5. Contribution ledger — 2 hours to build initial, maintained 10 min/quarter thereafter
  6. Miroslava classification memo from JME / Portilla — a $500-1500 ask, eliminates labor-reclassification risk
  7. Cindy Canto memo to file — 20 min of Austin's time, produces a defense document
  8. Transaction memos for the >$100K one-offs — 2 hours, closes documentation gaps on ~15 transactions
  9. Xero missing-tags pre-check — configurable once, runs forever
  10. Purpose-description standard for Xero — convention change, zero marginal cost per transaction
  • accounting/vendor-mappings.md — the counterparty patterns this memo references
  • accounting/chart-of-accounts.md — dimension structure and account plan
  • accounting/personal-card-business-expenses.md — specifically on personal-business commingling
  • accounting/casa-moksha-integration-plan.md — the RHG-side work sequencing
  • open-questions.md — the live counterparty-identification tickets
  • todo/for-austin.md — where the action items above would land