Case study

From a full day to 2 hours per loan.

How a $500m corporate lender automated portfolio monitoring, saving $180K+ a year in analyst time while delivering deeper risk analysis.

The challenges behind every monitoring cycle.

Inconsistent input formatsBorrowers report across PDF, PowerPoint, Excel and accounting systems, all structured differently. Each month starts with figuring out what has arrived and where the data sits.
Manual data entryFigures are manually transferred from borrower reporting into the fund’s model. It’s slow, prone to error and breaks when reporting formats or accounts change.
Limited trend analysisTime spent moving data leaves little for analysing it. Covenants, correspondence, performance and market conditions are reviewed separately, so trends and connections across months or sources are easy to miss.
Knowledge stays siloedWhat was checked, why decisions were made and what to watch next often lives with the individual. When ownership changes, that context is easily lost.
Days lost to every reportEach monitoring report takes an analyst around a day, most of it mechanical. That time could go to deeper diligence, structuring and origination, where judgement actually matters.

What did we put in place?

AI agents that complete the monitoring work, a dashboard that shows what happened, and a record of how.

The loan monitoring agents.

Agents that automate each step of the monitoring cycle inside the fund's existing AI tool.

  • Collect the reporting and chase what is late
  • Map the financials into the monitoring model
  • Test every covenant against your own definitions
  • Flag what moved and draft the note for sign-off

The monitoring dashboard.

Every loan's monitoring cycle on one screen, building a record of trends and risks.

  • What was received and what the analysis found
  • Cross-period trends for each borrower
  • Late, due and outstanding reporting
  • Last cycle's concerns, and how they have changed
  • Deeper insight from the reporting, for everyone

The audit and compliance portal.

A full record of every monitoring run, every action and every output, its sources, decisions and approvals.

  • What each run read, calculated and produced
  • Every figure linked to its source document
  • Versioned instructions and covenant definitions
  • Analyst sign-off on every note

What did the rollout look like?

Live in 4 weeks

From embedding with the team to live monitoring across the whole book, with the agents tested against past monitoring periods before they went live.

Embedded to live across the whole book.

10 analysts onboarded

Each analyst trained to run the agents from Claude, review what they produce and sign off every monitoring note before it is filed.

Running and signing off the agents in Claude.

30 loans monitored

Every loan in the book runs through the same agents each cycle, with its own context file holding its reporting, covenants and cadence.

Every loan on the same agents, every cycle.

Three credit categories

Agriculture, NBFI and corporate borrowers, each with its own reporting and covenants, handled on one monitoring process.

Three borrower types, one monitoring process.

Fund-wide visibility and accountability

Monitoring becomes a shared record, so everyone can see what’s been checked, what needs attention and who’s responsible for following up.

One record of every check, and who owns it.

The quantified ROI for this fund.

Time backSignificant analyst time back, spent on higher-impact work.
A system of workEvery monitoring run stored, so nothing lives in one analyst’s head.
ScaleWrite more loans and grow funds under management with the same team.

Assumptions

Average analyst salary
$150,000 a year
Working days per year
230
Working day
9 hours
Loans monitored
30
BeforeNow
Monitoring time per loan1 day2 hours
Monitoring cost per loan$652$145
Loans monitored per cycle3030
Total monitoring time per cycle30 days6.67 days
Cost of a full monitoring cycle$19,570$4,350
Annualised saving for this fund$182,640$15,220 a cycle, over twelve monthly cycles
See the full breakdown of the calculations
Day rate
$150,000 ÷ 230 working days
$652
Hour rate
$652 ÷ 9 hours
$72
Before
30 loans × 1 day × $652
$19,570
Now
30 loans × 2 hours × $72
$4,350
Saving a cycle
$19,570 − $4,350
$15,220
Saving a year
$15,220 × 12 monthly cycles
$182,640

Stay ahead of risks with
AI-powered loan monitoring.