What is Claude office building underwriting after the return-to-office shift? Claude office underwriting is the use of Anthropic's Claude Opus 4.7 model to underwrite a Class A or Class B office acquisition by integrating in-place lease economics, market sublease availability, hybrid-work utilization data, tenant credit health, and capital expenditure requirements into a single, defensible underwriting model. In 2026, with the return-to-office (RTO) shift creating bifurcation between trophy assets and commodity space, traditional spreadsheet underwriting tends to either over-credit RTO momentum or over-discount risk. Claude lets the analyst stress-test both sides at the same speed. For broader context, see our guide on AI multifamily underwriting, which uses many of the same Claude workflow patterns.
Key Takeaways
- Office underwriting in 2026 must integrate four signals that did not matter in 2019: hybrid utilization rates by submarket, shadow sublease inventory, tenant credit migration, and capex required for amenity competition.
- Claude Opus 4.7 can ingest the rent roll, T12 operating statements, lease abstracts, and submarket sublease report in one prompt and return a stress-tested 5-year proforma in roughly 15 to 25 minutes.
- The most consequential single assumption in post-RTO office underwriting is renewal probability by tenant cohort, which Claude can model conditional on tenant industry, lease maturity, and submarket utilization.
- Always run a downside scenario where the largest tenant does not renew and the space is re-leased at submarket asking rent net of free rent and tenant improvements, then compare DSCR and equity multiple.
- Office cap rates have widened materially in commodity Class B in many markets while compressing in trophy Class A, which means a single market cap rate is rarely the right exit assumption.
Office Building Underwriting in the Post-RTO Era Explained
Office leasing data through Q1 2026 confirms what most CRE professionals see in the field: a sharp bifurcation. Trophy Class A in CBDs with strong amenity bases and transit access is leasing at or above pre-pandemic rents; commodity Class B in suburban office parks is trading at 30 to 50 percent discounts to 2019 valuations and clearing the market through conversion to multifamily, life science, or simply demolition. Cushman & Wakefield's 2026 outlook confirms continued bifurcation, and CBRE has published submarket-level utilization data that quantifies the gap.
The implication for underwriting is direct. The 2019 playbook (mark-to-market the rent roll, apply a market cap rate, capitalize CapEx) systematically fails in 2026 because it assumes a single market clearing rent and a single market cap rate. Both have fragmented. Claude Opus 4.7's ability to ingest a 1 million token deal package and reason across the lease, market, and tenant data simultaneously makes the multi-signal underwriting tractable in a single sitting.
The Four Post-RTO Underwriting Signals
Signal 1: Hybrid Utilization by Submarket
Submarket-level utilization data (typically reported as percentage of pre-pandemic average occupancy or as average days-per-week in office) drives the renewal economics. A submarket where Tuesday-through-Thursday utilization is at 70 percent of pre-pandemic average will support different lease economics than a submarket at 35 percent. Pull this data from Kastle, Placer.ai, or local broker reports and feed it to Claude as a structured input.
Signal 2: Shadow Sublease Inventory
Direct asking rent does not clear the market when a submarket has 18 to 24 months of sublease inventory at 30 to 40 percent discounts. Always pull the most recent submarket report and have Claude calculate the effective competitive rent (weighted average of direct asking and sublease asking) for any space your underwriting assumes will be leased.
Signal 3: Tenant Credit Migration
The credit profile of office tenants has shifted. Some industries (AI infrastructure, biotech, financial services) are expanding office footprints; others (legacy media, traditional consulting back-office) are shedding. Have Claude classify each in-place tenant by industry and rate the likelihood of renewal at maturity using a simple three-tier system: high-renewal, neutral, low-renewal.
Signal 4: Amenity Capex
The flight-to-quality dynamic means commodity office must invest in amenities (fitness, food and beverage, conferencing, lobby modernization) just to hold market share. Underwrite a per-square-foot amenity capex line, typically $25 to $75 per SF on top of TI, and confirm DSCR holds.
The Claude Office Underwriting Workflow
The end-to-end workflow uses three structured prompts. CRE investors who want hands-on workflow setup can connect with Avi Hacker, J.D. at The AI Consulting Network.
- Lease-by-Lease Renewal Probability: Upload the rent roll, lease abstracts, and tenant credit signals. Ask Claude to assign each tenant a renewal probability (0 to 100 percent) at lease maturity, with a one-sentence rationale per tenant referencing tenant industry, lease maturity year, and submarket utilization.
- Cash Flow Construction: Provide the in-place NOI, the operating expense ratio, the submarket sublease overhang, the amenity capex assumption, and the projected re-leasing rent. Ask Claude to construct a 5-year proforma cash flow with two scenarios: base case (renewal probabilities from Step 1) and downside (largest tenant does not renew, space re-leases at submarket asking rent net of free rent and TI). Validate the math against your underwriting workbook.
- Returns and Sensitivity: Compute IRR, equity multiple, year-1 cash-on-cash, year-1 DSCR, and stabilized DSCR for both scenarios. Then ask Claude to identify the three input assumptions that most move IRR and to run a tornado sensitivity. The deal-or-no-deal decision usually turns on one or two of these three.
For deeper context on integrating Claude with rent roll automation, see our step-by-step guide on how to automate rent roll analysis with Claude Projects.
The Single Most Consequential Assumption
In post-RTO office underwriting, the single largest IRR driver is renewal probability on the largest 3 to 5 tenants. A 70 percent average renewal assumption versus a 50 percent assumption can swing the unlevered IRR by 400 to 600 basis points on a 5-year hold. The discipline is to make the renewal assumption a function of observable data (tenant industry, lease maturity, submarket utilization, in-place rent vs market) rather than a wish. Claude is well suited to this kind of conditional probability reasoning when the inputs are explicit.
The Capital Stack Implication
Lenders in 2026 are underwriting office at materially tighter DSCR and lower LTV than they were in 2021. A typical Class B office acquisition that would have cleared at 65 percent LTV with a 1.25x stabilized DSCR in 2021 may now require 50 to 55 percent LTV with a 1.40x to 1.50x DSCR. (Note: LTV is loan amount divided by appraised value, expressed as a percentage; DSCR is NOI divided by annual debt service, a ratio above 1.0 means income covers debt.) Have Claude validate that the capital stack you are modeling is achievable at the lender's terms by running the underwriting at the lender's cap rate, NOI, and debt-service assumptions, not yours.
Frequently Asked Questions
Q: Can Claude pull submarket utilization data on its own?
A: No. Claude does not have real-time data access by default. You must provide the utilization data as part of the prompt input. Pull from Kastle, Placer.ai, or your broker's submarket report and paste in.
Q: How does Claude handle conflicting data (for example, broker says 70 percent utilization, Kastle says 45 percent)?
A: Tell Claude in the prompt how to resolve conflicts. The standard approach is to use Kastle as the primary source (badge-swipe data is closer to ground truth than broker estimates), with broker data as a triangulation point. Always document the assumption in the underwriting memo.
Q: Should I trust Claude's renewal probability output?
A: Trust the structure, validate the substance. Claude's tenant-by-tenant probabilities are a useful first cut, but the deal team's domain knowledge of specific tenants (a CFO leaving, a parent company restructuring, an industry shift you have inside info on) should override Claude where you have it.
Q: What about office-to-multifamily conversion as a downside?
A: For commodity Class B in markets where conversion is feasible, run a third scenario: vacant building, conversion to multifamily, and exit at a multifamily cap rate net of conversion cost. Claude can model this with the help of construction cost benchmarks and submarket multifamily comps; just provide the inputs.
Q: How does this Claude workflow compare to a traditional Argus model?
A: Argus remains the institutional standard for tenant-by-tenant cash flow projection and sensitivity. Claude is complementary: use it for first-pass screening, downside scenarios, and narrative output for memos. For an institutional acquisition, the final-stage model still typically lives in Argus.