Case studies
Five industries, one pattern: targeted automation removes labor and leakage fast enough to show up in the P&L the same year.
Watch the platform run live.
Full logistics automation — inbound orders, dispatch, documentation, and invoicing — driven entirely by Poel AI.
Exception queues ate margin until routing went autonomous.
A mid-market 3PL was burning roughly 1,900 agent-hours per month on shipment exceptions: address fixes, carrier API mismatches, and dock scheduling conflicts. Supervisors were approving the same resolution paths hundreds of times, and expedited freight was triggered late because nobody saw the pattern until end-of-week.
We deployed an exception router that normalizes carrier payloads, applies business rules, and only surfaces true edge cases. Exceptions that used to need two humans now close in one automated pass; the rest queue with a suggested fix and a one-click approve.
Year-one impact: they avoided three net-new coordinator hires ($210k loaded cost), cut expedited freight spend by 18% on affected lanes, and recovered about $640k in fully loaded labor and fees. Build plus run cost landed under $180k for the first year, so they cleared payback in a little over four months.
We stopped paying people to babysit the same ten exception types. The system resolves them before they hit a supervisor inbox.
— Michael S., VP Operations, North American 3PL
KYC refresh and data reconciliation stopped being a quarterly crisis.
A regional wealth and lending group ran client refresh, sanctions screening, and CRM updates through a patchwork of spreadsheets and offshore overnight batches. Every quarter close meant 40–50 people in a war room reconciling mismatched customer records between core banking, CRM, and the compliance archive.
Automation now pulls golden-record rules across systems, runs continuous delta checks, and generates exception packets with source-line evidence. Analysts work exceptions, not empty hunting for what broke.
They retired two overlapping SaaS workflow tools ($94k annual) and redeployed roughly 11 FTEs of manual reconciliation effort into client-facing work. All-in, they booked about $780k in first-year operational savings versus $220k for implementation and managed run — roughly a 3.5x first-year ROI on cash outlay.
Quarter close used to mean fifty people proving the same numbers. Now the numbers reconcile themselves and we argue about strategy, not spreadsheets.
— Sarah B., Head of Operations, Financial Services Group
Prior auth and intake prep cut denials and rework before clinics opened.
A multi-site specialty network was losing almost $3M annually to prior-authorization denials and incomplete intake packets. Nurses and front-desk staff retyped the same clinical facts into three systems, and auth coordinators started their day triaging PDFs instead of submitting clean cases.
We wired structured intake, payer rule packs, and clinical snippet extraction into a single prep pipeline. By the time a human touches a case, payer-required fields are complete, attachments are labeled, and the submission path is chosen automatically.
Denied auth rework dropped 31% in nine months. They recovered roughly $2.1M in previously written-off or delayed revenue, while cutting coordinator overtime and weekend catch-up shifts. Implementation was about $290k; they modeled payback at roughly six months on revenue acceleration alone.
Patients used to wait while we hunted for a missing document. Now the packet is complete before the coordinator even logs in.
— David L., Chief Operating Officer, Specialty Network
Support routing and refunds stopped leaking margin on autopilot mistakes.
A high-volume DTC brand had scaled support with offshore tiers, but refund approvals, warranty lookups, and VIP routing were inconsistent. Agents were either over-refunding to keep CSAT high or stalling tickets while they asked managers — both expensive.
We shipped a policy engine tied to order history, loyalty tier, and inventory position. Tier-one resolves refunds, replacements, and routing with guardrails; anything outside policy escalates with a full context bundle for a manager.
They shaved average handle time by 22%, reduced refund leakage by about $420k in the first year, and avoided adding a planned 25-seat BPO extension (~$300k). First-year automation spend was near $165k, so total cash benefit versus spend was better than 4:1.
We still say yes to customers — we just don't say yes with someone else's credit card by accident anymore.
— Elena K., Director of CX, DTC Brand
Manual ledger and dispute ops were hiding eight figures of trapped efficiency.
A payments company processing tens of millions of transactions monthly still closed sub-ledger exceptions with a rotating offshore team. Dispute evidence was copied between ticketing, risk, and finance tools, and month-end accruals slipped when reconciliation queues backed up.
We automated ingestion of processor files, matched disputes to ledger lines, and generated finance-ready adjustment packets. Risk and finance now review exceptions with machine-prepared evidence instead of rebuilding the story from screenshots.
They freed roughly 35 FTE-equivalent hours per day across ops and finance, retired a chunk of outsourced reconciliation spend, and redirected internal capacity toward expansion markets. Finance leadership attributed about $1.2M in annual run-rate savings to the first wave of deployments, with a nine-month payback on the program investment.
We finally stopped paying smart people to be human copy machines between systems.
— Robert T., VP Finance, Series C FinTech