— Operations & Process
Assesses a refund or compensation request against the contract terms and refund policy that govern it, then returns a decision on approving it in full, in part, offering a goodwill gesture or declining, with every element of the remedy traced to the clause that allows it and anything beyond an agent's authority routed for review.
1–2 hours → ~5 minutes
For one document, brief or record at a time
No coding required
— USE CASES
Assesses contractual entitlement, service-failure evidence, customer history and compliance with claim windows and notice requirements. Provides an approval, partial refund, goodwill gesture or decline recommendation supported by the relevant terms.
Calculates compensation or service credits using the applicable remedy schedule. Connects each proposed amount to the clause and formula that support it, making the recommendation easier to explain and review.
Identifies proposed remedies above the handling agent’s authority limit and flags them for review. Keeps the financial recommendation distinct from permission to pay, with authoriser approval required for every payment.
Flags incomplete evidence, open chargebacks, legal claims and requests outside the permitted window or without a contractual or policy basis. Records the issues that prevent a straightforward approval.
Refund and compensation requests create difficult judgement calls when policy wording, contractual rights, service failures and customer expectations do not align. In practice it shows up as disputed refund eligibility: assesses contractual entitlement, service-failure evidence, customer history and compliance with claim windows and notice requirements. Provides an approval, partial refund, goodwill gesture or decline recommendation supported by the relevant terms. The value sits in the rigour, not the typing — yet the rigour is exactly what gets traded away when there is only 1–2 hours of capacity for it.
Skillsize turns that work into a Skill: you supply the material, and what comes back is approval, partial refund, goodwill gesture or decline recommendation, with clause-grounded decision record with supporting quotations. What sits between input and output is the codified method: thresholds, sequencing and the points where a human confirms a call — all of it visible and editable in the Skill. In effect, 1–2 hours of senior time compresses into ~5 minutes — and the output is comparable across clients, quarters and colleagues instead of shaped by whoever ran it.
Refund & Compensation Engine exports as a structured SKILL.md file and is MCP-ready, so the same method runs in ChatGPT, Claude, Copilot or your own AI products. Adapt it to your methodology, and the intelligence stays yours — not locked to one vendor.
Identifies where adoption of a new system, process or way of working stalls after go-live, with an evidence-backed report on barriers, workarounds and differences between functions.
Assesses AI readiness through anonymous staff evidence, identifying where AI can improve day-to-day work and the practical barriers that could prevent those gains.
Maps how an organisation actually operates using anonymous accounts from its people, producing an evidence-backed diagnostic of responsibilities, decisions, handoffs and duplicated effort.
Reads a set of weekly status notes for progress, blockers and RAG, charts the trend, and writes an executive digest focused on trajectory.
Reviews a proposed change against your change policy and returns a governed advisory decision — approve, approve with conditions, defer to the change board or reject — weighing risk, blast radius and rollback, with a change manager sign-off required before anything reaches production.
Weighs an open case against your escalation policy, its risk and its commercial exposure, then returns a governed recommendation to hold, monitor, escalate or take it to executive level, with the reasoning shown and an owner sign-off required — nothing is escalated automatically.
