Consumer lending and pawn / SRE / on-call engineer / Incident replay
03:02 to 03:09 at F88
One production incident, replayed minute by minute, with the approval tap included.
Published July 20, 2026/01 / 06
Root cause used to take days at F88: six dashboards, three log systems, one engineer reconstructing a timeline by hand. Here is what the same class of incident looks like now, minute by minute.
The alert fires at 03:02. By 03:04 the investigation is already written: signals correlated across the observability stack, the suspect change identified, a remediation drafted. At 03:05 the on-call engineer gets one message with the evidence and one question: approve or hold. The fix is applied and verified by 03:09. The engineer read a summary, made one decision, and went back to sleep.
The point is not speed for its own sake. Every step of those seven minutes is in the audit log: what was checked, what was found, who approved. That is the difference between automation a regulated lender can run and automation it has to ban.
#AISRE #FinServ #Resolve #CloudThinker
02 / 06 The problem
The 03:02 alert, the old way: an hour of tabs before the first theory
03:02
Page fires
Latency alert on a core service. On-call wakes up, opens the laptop.
03:10
Tab sprawl
Metrics in one tool, logs in another, traces in a third, last night’s deploys in a fourth.
03:45
First theory
Maybe the replica. Maybe the deploy. Nobody is sure enough to act on a regulated system.
04:30
Second engineer paged
The one who knows this subsystem. Now two people are awake.
06:00
Mitigated, not explained
Service restarted. Root cause still unknown, so it will happen again.
Day 3
The report
Root cause finally written up, from memory, for the incident review.
Days of elapsed time, for a cause that was visible in the data at 03:02.
03 / 06 Why the current answers fail
Two ways to fail: keep doing it by hand, or hand production to an agent on day one
Option A: the status quo
One-off scripts
one per account, none the same
Tribal knowledge
the standard lives in someone’s head
Manual follow-up
tickets chased, dashboards watched
Quarterly evidence
compliant in audit week, unknown after
Resultsafe, does not scale
Option B: AI with full access
Trust on a promise
a demo is not a control
Unscoped credentials
one key for cloud and on-prem alike
Recommendations only
still a human acts, still no evidence
No audit trail
the regulator asks; the chat log answers
Resultscales, nobody signs off
What F88 chose
Neither. Agents investigate every alert and draft the fix; a person approves anything that touches production. The 03:02 page becomes one decision, not one shift.
Trust is not granted to a system. It is earned, one audited check at a time.
04 / 06 Frontier Investigation Agents
Four steps from read-only to autonomous. The team decides when to take each one.
Step 1
VerifyRead only
Live inventory, drift, waste and health checks. Nothing changes. Every finding explains what was checked, what was found, why it matters.
Step 2
EncodeStandards as Skills
Naming, tagging, guardrails, cost controls and runbooks written once and run the same way in every account. Tribal knowledge becomes a file.
Step 3
The gateAct with approval
Agents investigate and propose. A person approves where risk, policy or production impact demands it. Every action is written to the audit trail.
Step 4
Auto modeAutonomous under guardrails
Per environment, the team sets Notify, Act-with-Approval or Autonomous. Low-risk remediations run on their own. Production keeps the gate.
The on-call engineer’s question
“Will it act while I sleep?”
Only inside the autonomy level your team set for that environment. Low-risk checks run on their own. Production changes wait for the approval tap, and every step is in the log either way.
Frontier Investigation Agents for your cloud. They scale the team’s judgment, they do not replace it.
05 / 06 Proof
The same incident class, after: 03:02 to 03:09
03:02
Alert fires
Resolve picks it up and starts correlating signals immediately.
03:04
Investigation written
Root cause identified across metrics, logs and the deploy timeline. Remediation drafted.
03:05
One message
On-call gets the evidence and one question: approve or hold.
03:09
Fixed and verified
Change applied, metrics confirmed back to baseline, audit log complete.
Days to root-cause
Minutes
Resolve correlates signals across the observability stack and hands the team a structured summary.
Daily ops by hand
80%
of daily manual operations tasks automated, human approval where risk demands it.
The engineer made one decision. The system did the reconstruction.
06 / 06 One number, and the ask
The only number we ask you to check yourself
Read-only, first 30 days
Zero changes to production
Start the way our customers did. One scoped, read-only key. The engine inventories your estate and reports drift, waste and health. Your team checks the findings against what it already knows. If they hold, you decide the next step.
What you get
1 report
Full inventory, drift and waste findings, each with what was checked and why it matters.
What you risk
1 read key
Scoped and revocable. Every check it runs is in the audit log from day one.
Do not trust the AI. Verify it. Then let it earn the next step.
Pilot shape as run at F88.
Sources
- F88 (Consumer finance): 80% of daily ops automated, AWS spend down 30%, continuous compliance.
- Incident pattern and figures: F88 case study
The evidence
F88
80% of daily ops automated, AWS spend down 30%, continuous compliance.
Read the full evidenceThe module
Resolve
The part of CloudThinker this story runs on, in product terms.
Explore Resolve