Free Strategy Call

Book Your Call

One call. Clear direction. Faster execution.

Case StudyOps·AI Agents

How a distributor automated 9,000 hours of back-office work with AI.

Finance, HR and operations teams were spending their days copying data from one system to another. Skillfi's fractional CIO mapped every manual process, picked the ones worth automating and launched AI agents with people approving every important step. It's how our fractional CIO service works: technology that makes the whole business run better.

Yearly savings
$0→$1.8M
9,000Hours automated a year
8AI agents live
−91%Data entry errors
01

Skilled people working as human copy-and-paste.

The company distributes industrial parts from five warehouses, running on nine systems that didn't talk to each other: an ERP, a CRM, payroll, an HR system, a warehouse system and several spreadsheets.

Finance keyed supplier invoices by hand. HR typed new-hire details into four systems. Operations copied order changes between the CRM and the warehouse system. Errors caused late payments, payroll mistakes and wrong shipments, and every team kept asking for more headcount. Leadership wanted to use AI, but had no plan and worried about letting it touch financial data.

9disconnected systems
60manual processes found
4.2%data entry error rate

We were hiring accountants and HR people, then asking them to spend half their day typing.

CFO, client
02

Automate what's worth it, keep people in charge of what matters.

Our fractional CIO started with the work, not the technology, then chose where AI agents would pay back fastest and safely. Teams learned to work alongside the agents through our Claude workshops.

Map

Every manual process measured

60 processes across finance, HR and operations mapped, timed and scored on hours saved, error risk and how easy they were to automate.

Automate

Eight AI agents for the biggest wins

Agents that read invoices, update records across systems, check data and prepare work, connected to the existing systems instead of replacing them.

Govern

People approve what matters

Human approval for payments and pay changes, clear limits on what each agent can do, full audit logs and a weekly exceptions review.

The Engagement

6 months, step by step

From copy-and-paste work to AI agents with human oversight. Here's how the engagement unfolded.

  1. Month 1

    Process discovery

    Workshops with every team to map, time and score their manual processes.

    60 processes
  2. Month 1–2

    Business case & guardrails

    Top 8 processes chosen, savings estimated and rules set for data access, approvals and audit trails.

    Top 8 chosen
  3. Month 2–3

    Finance agents

    Invoice processing and payment matching agents built, tested and run alongside the team first.

    2 agents live
  4. Month 3–4

    HR & operations agents

    New-hire setup, order updates, stock checks and supplier follow-ups automated across systems.

    8 agents live
  5. Month 5

    Team training

    Every affected team trained to review agent work, handle exceptions and suggest new automations.

    140 people trained
  6. Month 6

    Results & roadmap

    Savings measured and a two-year automation roadmap presented to the board.

    Board approved
Inside the AI Agents

Agents across the back office

Eight agents working across finance, HR and operations, with clear limits and a person approving anything that matters.

Finance3 agents

Invoices to payment

  • Invoices read & coded
  • Payments matched
  • Approval before paying
HR2 agents

People data in sync

  • New hires set up once
  • Changes flow to payroll
  • Pay changes approved
Operations3 agents

Orders & suppliers

  • Order changes synced
  • Stock checks
  • Supplier follow-ups
ControlAlways on

Safe by design

  • Limits on every agent
  • Full audit logs
  • Weekly exceptions review
Every month: Automation reviewHours saved, accuracy and exceptions reviewed with each team every month, and the next processes chosen from the roadmap.
Month 1 · 60 processes mapped Month 4 · 8 agents live Month 6 · $1.8M a year
The Results

Six months after kickoff

Time to process a supplier invoice

Keyed by hand14 min
AI agent + human check2 min
051015

Across 3,000 supplier invoices a month.

Hours automated a year—→9,000
Yearly savings—→$1.8M
Data entry error rate4.2%→0.4%
New back-office hires needed6 planned→0
AI incidents—→0
CF

Our teams now review work instead of typing it. We saved $1.8M without a single layoff, and every payment still gets a human sign-off.

CFO, client
Your Turn

Are your teams copying data between systems?

Our fractional CIO will map your manual work, choose where AI agents pay back fastest and launch them safely with your people in control. Want your teams confident working with AI? Add a Claude workshop.

What we'd build for you
  • A manual process map & business case
  • AI agents for your biggest wins
  • Guardrails, approvals & audit logs
  • An automation roadmap
Book a Strategy Call

30 minutes. No obligation.

Let’s Work Together

Contact Us

Sit back, relax. We will get back to you soon.

Related Case Studies