Written from scratch, every time
An analyst pulls financials, re-types the same spreads, and rebuilds a memo that looks like the last forty. The judgment takes an hour. The assembly takes a day.
AI modernization for oilfield, service, and fintech companies still running critical workflows by hand.
We start where the work actually happens — the spreadsheet everyone edits, the approval that waits on one inbox — and modernize the loop around your analysts, not instead of them.
Book an audit
A two-week look at one workflow.
The situation
Critical workflows still run by hand.
Not because anyone chose it. The system was built for a smaller company, the person who understood it left, and the workaround became the process. Everything below is work we have watched competent teams do manually, every month, on a deadline.
Keyed in twice, sometimes three times
Written in the field, photographed, emailed, re-entered into the billing system. Every hop is a place a number can change without anyone noticing.
Reconciled against a PDF
Joint interest billing arrives as statements that do not match the ledger. Someone opens both, scrolls, and marks the differences by eye before the close.
Waiting on one person's inbox
Limits, exceptions, releases — all real decisions that need a human. But the queue lives in email, so nobody can see what is stuck or how long it has been stuck.
The method
Map it, modernize it, keep people in it.
Map the workflow
We sit with the people doing the work and trace one process end to end — every hand-off, every re-key, every wait. Nothing is proposed until the map is right.
Modernize the loop
Automate the assembly, not the judgment. Documents parsed, data reconciled, drafts prepared — so the analyst opens something finished enough to argue with.
Keep humans in it
Every consequential step keeps a named reviewer, a visible queue, and a record of what changed. Your analysts, faster and surer.
We map the workflow before we change any of it.
The hearth was never the flame. It was the architecture that kept it.
Placeholder frame — graded fire footage pending
What an audit surfaces
Three findings, written plainly.
An audit ends in a short document: where the time goes, what can be modernized in ninety days, and what should be left alone. The examples below are illustrations of the format — not client results.
The memo is 80% assembly
Of the average credit memo, most of the elapsed time is gathering and formatting. The analytical section is written last, tired, and quickly.
● assembly 6.5 hrs · judgment 1.2 hrs
Illustrative figures. Not a client result.
The same number is entered four times
A field ticket quantity travels from paper to photo to email to the billing system. Each re-entry is unverified, and the mismatch is found at invoicing.
● 4 entry points · 1 reconciliation, at month end
Illustrative figures. Not a client result.
Nobody can see the queue
Approvals sit in individual inboxes. There is no shared view of what is waiting, who owns it, or how long it has been there — so escalation is a guess.
● 3 approvers · no shared queue · age unknown
Illustrative figures. Not a client result.
Who we are
A small practice, run by someone who has done the work.
Hearth Advisory is a credit and operations practice in Houston. The background is portfolio risk — writing the memos, chasing the exceptions, closing the month — which is why the work starts with the workflow and not with the technology.
We take a small number of engagements at a time, we say when something should not be automated, and we leave documentation your team can run without us.
Start here
The AI Modernization Audit.
Two weeks. One workflow, mapped end to end. You get the map, a ninety-day modernization plan, and a plain answer about what is worth doing. If the answer is “less than you think,” we will say so.
Book an audit
Or write plainly about the workflow that keeps you up at month end.