Price · Deploy · Expand · Govern is how the operator experiences the business. No underwriter has "loop coverage" in their risk model. An underwriter thinks in four buckets: can they service the debt, what can I lien, who pays them, and what breaks under stress. This page is the translation — the same vault, projected into the 16 fields a lender's covenant structure actually uses.
▸ Capture broad, translate narrow. The 16 modules are the capture layer. The Pulse Engine is the translation layer. The lender never consumes "16 modules" — they consume their own underwriting inputs, pre-filled and attested.
The operator organizes around how work flows. The underwriter organizes around how risk is structured. The same captured evidence has to be projected into the second model — because that's the only one a credit committee can act on.
This is how the operator runs the business day-to-day. It's the right shape for capturing evidence — and the wrong shape for handing to a credit committee.
This is the model the lender already uses. Allometry's job isn't to teach them a new one — it's to pre-fill the one they have, with verified, attested values.
▸ 16 modules → 16 fields
Every operator module maps to a specific, named field in a lender's risk model — grouped under the four buckets a credit committee thinks in. Each field shows what it answers, which module produces it, and its honest data lineage: native (Allometry generates it), partner-augmented (needs an integration), or constructed (cohort + external).
Is there enough win-weighted forward revenue to cover scheduled debt service?
How much revenue is already signed, and for how many months on weighted average?
How much of deployable capacity is already committed — is growth real or theoretical?
What's the blended margin, and does the floor hold quarter over quarter?
What receivables exist, how old, how much is borrowing-base-eligible after dilution? The core ABL field.
On-hand inventory worth, turn rate, obsolescence and stockout exposure.
What equipment is deployed, where, and what's the depreciated net book value?
How much cost is loaded into in-flight jobs not yet billed — and at what completion?
How dependent is revenue on a few customers — what's the single-customer exposure?
Payment terms, auto-renew, and how easily can customers walk (term-for-convenience)?
Do customers pay on time — days-to-pay by account — and which are flagged at-risk?
What industries do the paying customers sit in — how correlated is that to a downturn?
Does the operator deliver reliably — and what breaks if a key person or site goes down?
Could this operator pass an audit today — where are the open control gaps?
Can the lender trust the data itself — is every field tamper-evident and hash-chained?
In a stress quarter, can the operator originate enough new revenue to backfill losses?
▸ covenant → module → trigger
A field is not a covenant. A covenant is a field plus a threshold plus a breach condition. Below: every covenant an institutional facility runs on, the Allometry module that instruments it, and the trigger value that flips it to default. This is the translation layer between what the lender enforces and what the model already measures.
| Covenant — what the lender enforces | Module(s) | Breach trigger | Status |
|---|---|---|---|
| Customer concentration capno single customer above 20–25% of TTM revenue | M.11 · M.12 | Top-1 customer share of trailing-12-month revenue crosses the cap | Native today |
| Backlog / forward-revenue coveragecontracted backlog ≥ 1.0–1.5× next-12-mo debt service | M.01 · M.04 · M.09 | Contracted forward revenue ÷ scheduled debt service falls below the coverage multiple | Native today |
| Gross margin floortrailing margin ≥ operator-specific floor set at close | M.02 · M.03 · M.06 | Rolling 90-day realized gross margin crosses below the floor | Native today |
| SLA / uptime complianceO&M and PPA-backed facilities require uptime ≥ 95–97% | M.08 | Rolling SLA compliance rate drops below the contractual minimum | Native today |
| Revenue retention / churn floornet revenue retention ≥ covenant floor | M.11 | Cohort net revenue retention crosses below the floor | Native today |
| Supplier / input concentration capsingle-supplier input ≤ X% of COGS | M.03 · M.13 | Single-supplier share of COGS crosses the cap | Native today |
| Capacity / throughput maintenanceproject debt requires sustained throughput to hit milestones | M.07 | Forward capacity commitment vs. contracted obligation gap widens past tolerance | Native today |
| Compliance certificate / reporting cadencecovenant certificate delivered on schedule | M.14 · M.16 | Auto-generated from the WORM ledger — late filing is structurally impossible; a missing field is the only breach mode | Native today |
| Audit-readiness / data integrityclean, continuous audit trail | M.14 · M.16 | Evidence-completeness score below threshold; any hash-chain break | Native today |
| Material adverse change — early signalhistorically a qualitative "material deterioration" clause | M.16 | Composite Pulse score drops beyond a set quarter-over-quarter band — a quantified, continuous tripwire, not a post-hoc legal argument | Native today |
| DSCRcash available for debt service ÷ debt service ≥ 1.20–1.35× | M.03 · M.06 + Codat / Plaid | Computed DSCR crosses below the covenant floor | Partial — cash side Q4 2026 |
| Minimum liquiditymaintain ≥ $X cash or undrawn availability | Plaid / Flinks | Bank balance plus availability crosses below the floor | Needs Plaid / Flinks Q4 2026 |
| Maximum leveragetotal debt ÷ EBITDA ≤ the cap | Codat + M.03 / M.06 | Computed leverage ratio crosses above the cap | Needs Codat Q4 2026 |
The covenants Allometry instruments natively are not the residual set. They are the operational and performance covenants that most accurately predict default for field service operators — and the ones a traditional lender monitors worst. A lender today receives a quarterly certificate and trusts it. Allometry produces a continuous, attested reading.
The compliance certificate self-generates from the WORM ledger — late filing, one of the most common technical covenant breaches, becomes structurally impossible. And the material-adverse-change clause — historically litigated after the fact — becomes a quantified, continuous tripwire: the composite Pulse score, monitored against a set band.
DSCR, minimum liquidity, and maximum leverage are the pure balance-sheet covenants. Allometry contributes the margin and cash-generation side natively; the cash-position and debt-schedule side closes with Codat and Plaid / Flinks, sequenced for Q4 2026. Operational covenant coverage is live; financial covenant coverage is on a dated integration path.
Every row above resolves to a threshold — a number the operator must hold above or below. The zero-knowledge proof layer converts each threshold into a proof rather than a disclosure. The operator does not send the lender their customer list, their margin file, or their bank statement. They send a cryptographic proof that the covenant threshold is cleared — and the lender verifies it against the published methodology. Covenant monitoring stops being an event checked four times a year and becomes a state the lender holds continuously: a breach surfaces the day it occurs, not the quarter after.
A lender's underwriting process is, mechanically, the work of filling a model — gather the documents, normalize them, compute the covenant inputs, set the thresholds, decide. For a $500K facility that work takes six weeks, because the operator's evidence arrives as 35 unstructured documents and the analyst rebuilds the model from scratch. With the covenant → module → trigger map already in place, that work is already done — continuously, by the model, before the lender opens a file. The process collapses to three steps: verify the proofs, set the threshold values at close, decision. The operator didn't assemble a faster data room. The data room assembled itself.
▸ The output
Not the generic Pulse JSON. Not "16 modules." A covenant structure the lender's risk model can ingest directly — borrowing base, coverage, concentration, downside — every field attested.
The 16→16 mapping above is Allometry's best current model of what a Tier-1 underwriter needs. 12 of the 16 fields are native today; 3 are partner-augmented (they need a billing or accounting integration to be complete); 1 is constructed from cohort + external data.
What this page is not: a claim that we've validated every field with every lender. The covenant structure is strongest for Tier 1 — cash-flow lending against the operator. Tiers 2–4 (infra debt, insurance-linked, rated paper) underwrite assets and contracts, not operators — they need a different projection, and that's the expansion path, not today.
And telemetry supplements documents — it doesn't replace them. A signed security agreement is still a signed security agreement. The vault makes diligence fast and verifiable; it doesn't make the legal layer disappear.
The covenant → module → trigger map is built from first principles. The thresholds are the part only a Tier-1 underwriter can confirm — so we'd rather ship the version three lenders have pressure-tested than the version we modeled alone. 30 minutes — you get the sample profile and the written scan regardless.
16 evidence guardrails · 5 vintage layers · how the data gets captured and attested in the first place.
The agentic layer that runs the RAG → LLM → recurse → ZKP pipeline — and does the actual translating.
Free Portfolio Blind-Spot Scan · the sample profile, cohort sizing, and where this field list still needs your input.