Anmelden Demo anfragen
Solutions · Private Credit

Spot the problem before the test date

Private credit lives on maintenance covenants and borrower reporting that arrives late and in every format. ExactCov extracts the package at closing, reads whatever the borrower sends, recalculates on the agreement's own definitions and shows you headroom before it becomes a breach.

Demo anfragen Why maintenance is a calendar
Unitranche and senior facilitiesIntercreditor agreementsCompliance certificatesBorrower financials: PDF accounts, Excel packs, board decksAmendments and waiversSide letters
The problem

Maintenance covenants have a cadence. The spreadsheet does not.

The package is extracted at closing into a spreadsheet that drifts from the agreement's definitions two amendments later. Borrower financials arrive as PDFs, Excel packs and board decks, and someone retypes them. Headroom is known at the test date, not before, and the IC update is assembled by hand the night before.

CitedEvery ratio, threshold and step-down opens to its clause in the executed agreement, amendments read against the original.
DatedEvery test result carries the certificate it was run on and the date it arrived. Stale is a state, not an age.
EarlyHeadroom trends and warning bands flag the direction of travel weeks before the test date.
LoggedThe Breach and Waiver Log holds every decision with its history, ready for IC and LPs on any day.
The register

What is extracted from each facility

At closing, once. Then tested every period for the life of the facility.

Financial covenants

Each test as the agreement defines it, not as the template assumes.

  • Leverage, interest cover, cashflow cover, capex, liquidity
  • Step-downs, equity cure rights and springing tests
  • EBITDA add-backs and every definition resolved to its leaf

Reporting and deliverables

The calendar the agreement sets.

  • Compliance certificates, monthly and quarterly accounts, budgets
  • Due dates, grace periods and what counts as delivered
  • Chased through a secure link, never email attachments

Financial spreading

The borrower's numbers however they arrive.

  • PDF accounts, Excel packs and board decks read without templates
  • Ratios recalculated on the agreement's definitions
  • Each figure cited to the page it came from
The cadence

Maintenance covenants have a cadence. Incurrence does not.

Incurrence is tested when the borrower acts. Maintenance is tested every period whether or not anything happens, which is why one job is a calculation and the other is a calendar. A late certificate opens a stale zone that the book must show as stale, not as the last pass.

Read the piece
Q1Q2Q3Q4Q1 Incurrence "may they?" add-on: capacity?dividend: RP room?disposal: permitted? nothing to test between events, often for years Maintenance "are they, still?" cert duecert duecert duecert duecert due Q3 cert late: stale zone 4.1x, 18% room4.3x, 14% room4.6x, 8% room4.7x, 6% room the top line needs a capacity calculation on demand; the bottom line needs a calendar, a chase, a trend and a due date
Incurrence is tested when the borrower acts. Maintenance is tested every period whether or not anything happens, which is why one job is a calculation and the other is a calendar.
Three failure modes

Stale, late, missing: the three gaps nobody reports on

A certificate that arrived but is out of date, one that is overdue, and one that was never requested all produce the same symptom on the book: a pass. Only a calendar with due dates can tell them apart, and only a chase log can say who was asked and when.

Read the piece
JanFebMarAprMayJun today book shows Stale March figure still doing the work PASS tested in March, nothing since Late due not chased PASS due date passed, nobody asked Missing no calendar item was ever created PASS added to the umbrella, deliverable not set up official NAV received and tested
Three different gaps in the data, one identical symptom on the book. Only a calendar with due dates can tell them apart.
From our research

OCR for the documents borrowers actually send

Scanned execution versions, two-column schedules, footnoted thresholds and stamped pages. Recognition is one step of five. The gains came from knowing what kind of page it was before reading it, re-reading digits, and keeping the confidence so the hard 15 percent could go to a person, which is what lifts scans from 96.9 to 99.7 percent.

OCR best practices for credit agreements
Source checktext layer or scan?dictionary-hit rate Layoutcolumns, tables, notesstamps, headers dropped Recognisedeskew, 300 dpiboxes + confidence Reconciledigits re-read, totals addsections in sequence fieldconfidence Straight through85% of fields on scans Second read15%, ranked by risk Field accuracy on the 22 scans raw OCR, one engine88% + layout first, numbers re-read93% + confidence kept per word95% + reconciliation, first pass96.9% second read on the routed 15% takes scans to 99.7%; character error rate barely moved between rows two and four
Recognition is one step of five. The gains came from knowing what kind of page it was before reading it, from re-reading digits, and from keeping the confidence so the hard 15 percent could be sent to a person, which is what lifts 96.9 to 99.7.
Go deeper

Reading for private credit

Short pieces from the blog and the measurements behind them from Research.

Blog · 3 min

Maintenance covenants have a cadence. Incurrence does not.

Incurrence asks "may they?" on demand. Maintenance asks "are they, still, this quarter?" on a schedule. Why private credit monitoring is a calendar with a calculation attached, and syndicated tooling is the reverse.

Read Blog · 3 min

Stale, late, missing: the three failure modes nobody reports on

Most books can tell you a covenant passed. Very few can tell you what the test was run against, or that it has not been run since March. Days-stale is a first-class risk metric, not a data-ops footnote.

Read Blog · 3 min

The five things a credit exec should see before 9am

Due soon, overdue, blocked and by whom, near covenant, deteriorating fast. Five lists in a fixed order that turn the whole thesis into one screen, each row traceable to the documents behind it.

Read Blog · 3 min

What document extraction gets wrong

Threshold schedules, basket stacking, defined-term chains, grower baskets, restated sections and cross-references that leave the document. The six failure modes we test against, and what to ask a vendor to show.

Read Research · 4 min

OCR for credit agreements: what actually moves accuracy

Scanned execution versions, two-column schedules, footnoted thresholds and stamped pages. Nine practices that moved our first-pass field accuracy on scans from 88 to 96.9 percent, with the second read closing the rest, and the one metric that hides the damage.

Read Research · 4 min

How much human review is enough? Confidence routing and the second read

Reviewing every field is a second extraction. Reviewing none is a bet. Routing by calibrated confidence, a second read on 15 percent of fields takes our first pass from 97.4 to 99.7 percent. How to set the budget and what to do with the rest.

Read
On the platform

The parts private credit funds use most

Everything on this page runs on the same register as the rest of ExactCov. These are the parts this team lives in.

Platform

Command Center

The whole book on one screen: what has arrived, what is stale, what is in breach and what is drifting, with headroom trends, scenario shocks and committee packs built from the register.

Command Center Platform

Agents

Chase requests what is due on the calendar the documents set, Extract reads what lands, Test recomputes every affected covenant within nine seconds, Brief writes the morning list and Escalate treats silence as an event.

Agents Platform

Vault

Every agreement, amendment, certificate and notice, every version, with the page each number came from. Dependencies between documents tracked, regional residency and zero data retention.

Vault
Getting started

Run it in parallel. Don't run a pilot.

The documents are loaded once. What already arrives by email is copied to a mailbox we watch. Nothing changes for the team, and after one cycle the two outputs are compared, disagreement by disagreement.

Week 0Facilities loaded, package confirmed against the executed documents, reporting calendar generated.
Weeks 1 to 4Certificates and financials flow to both. Spreading, tests, chases and the morning brief run on our side only.
Week 4Compare: deliverables chased late, ratios that disagree, headroom the tracker did not show falling.
AfterPer facility in scope. IC and LP reporting from the register, on any day.

See it against your portfolio

Bring three facility agreements and the last two compliance certificates for each. We show the register, the headroom trend and what would have been flagged.

Demo anfragen Contact us