Make HR platform data do something employees can hold in their hands.
GiftWorks is the agentic merchandise and gifting layer for the employee lifecycle. Your platform stays the system of record. We operate the physical part underneath it.
Every company already knows when its employee moments happen. A hire. A work anniversary. A promotion. A recognition award. A retirement. That knowledge already sits in the HRIS, the human resources system of record, and in payroll, logged as a transaction long before anyone thinks about the experience around it.
What companies cannot do reliably is act on it. Welcome kits arrive in week three. Anniversaries get missed. Recognition turns into a gift card because nobody has time to do better. Every program is a spreadsheet, a stressed coordinator, and three vendors. The gap is not the data. The gap is the execution.
GiftWorks closes that gap. We detect the lifecycle event, apply the budget policy, route the approval, personalize the merchandise, coordinate the supplier, track the delivery, and report the outcome. Humans approve the rules. Bounded agents do the work.
We are not asking your platform to operate commerce, suppliers, or fulfillment. We are built to work with data from platforms like Workday and ADP, and to turn that data into something an HR buyer can see on a desk before day one.
Signal in. Governed. Delivered. Reported.
Why platform ecosystems
The one part of total rewards nobody owns.
Compensation has payroll. Benefits has benefits administration. Recognition has points software. Physical execution has a spreadsheet.
Total rewards is compensation, benefits, wellbeing, recognition, and development. Recognition and milestone programs are the fastest growing discretionary line in that stack, and the most operationally underserved. HR leaders describe the pain in the same three words every time: vendor sprawl, spreadsheets, chasing approvals.
That gap sits directly on top of your data and directly below your customer relationship. It is the reason a lifecycle event that your platform recorded perfectly still ends in a generic parcel four weeks late, or in nothing at all.
We operate that layer, and we do it without asking a platform to take on the parts that make it expensive.
We are not a swag catalog with an integration attached. The catalog is the least interesting part of what we do.
What the platform gains
- Lifecycle data becomes visible value. The Worker record, the job change, the continuous service date, and the eligibility flag stop being fields in a report and start being something an employee opens.
- No overlap with the core system. We do not hold the employee master, run payroll, or want to. We consume events and return outcomes.
- No commerce operations to build or staff. Suppliers, quotes, production, freight, exceptions, and returns stay with us.
- A purchasing motion your customers already recognize. Requirement, RFQ, which is the request for quote step where suppliers bid on a defined order, vendor pricing, purchase order, fulfillment, reporting.
- Governance your reviewers can read. Human gates, bounded agent authority, append only logs, minimal data.
Built for payroll platforms
Payroll holds the truth a gifting program depends on.
Tenure, eligibility, status, cost center. Get those wrong and the program is wrong, no matter how good the merchandise is.
A service award program is a policy problem before it is a merchandise problem. Who has reached five years of continuous service. Which cost center carries the spend. Whether the employee is active, on leave, or already in a termination workflow. Which entity and which location. Payroll systems hold that truth with more precision than any other source in the business, and they hold it continuously rather than in a quarterly extract.
That is why payroll-triggered programs behave differently from campaign-driven ones. The queue builds itself from real status. Tiers apply against the department and cost center that will actually be charged. Exceptions get flagged before an order is placed rather than after a finance review. Nobody rebuilds an eligibility list by hand every month.
GiftWorks consumes those signals with a deliberately small field set, applies the tier and budget policy the customer has approved, routes the approval, and reports spend back in the same cost center structure finance already uses.
Roadmap capability
The compliance loop, on our roadmap
Non-cash employee awards are generally treated as taxable fringe benefits in the United States, which creates imputed income, meaning the taxable value of a non-cash item that has to be reported through payroll. Today that reconciliation is manual, inconsistent, or quietly skipped. It is real work, it lands on the payroll team, and most merchandise vendors have never heard of it.
We are building toward a clean per-employee taxable-value feed emitted back into payroll, so that the award and its reporting close in one loop rather than two disconnected processes. This is roadmap capability, not a shipped feature, and we will describe it that way in every conversation. Thresholds, exclusions, and treatment vary and belong with each customer's tax counsel, not on a marketing page.
A layer that is integrated with payroll is the only kind of layer that can offer this. That is the specific reason payroll platforms should care about what we are building.
Talk to our partnerships team Download the platform overview
Built for HCM platforms
Lifecycle signals, turned into something the employee keeps.
A promotion is a business process event before it is a celebration. We handle everything after the event.
Human capital management platforms already model the employee journey with care. Offer accepted. Worker created. Job change. Continuous service date. Leave. Termination. Each of those is a structured event with an effective date, an owner, and an audit trail.
What is missing is the step after. The event fires, the record updates, and then a coordinator opens a spreadsheet. GiftWorks takes the event and runs the rest: policy, approval, employee choice, supplier, delivery, reporting. The HR buyer sees lifecycle data producing a visible result, which is the version of platform value that survives a renewal conversation.
We complement recognition systems of record
Recognition platforms such as Achievers, Workhuman, and Awardco decide who deserves what. They own points, nominations, and social feeds, and they should. We operate how it actually arrives.
That means we can sit underneath a recognition program rather than beside it. Points and moments awarded in those systems still have to become physical objects, sourced within budget, personalized, produced, shipped, and confirmed. We fulfill those outcomes. We do not ask a customer to replace the recognition system they already run, and we do not compete for that seat.
Procurement resonance
The flow from requirement to quote to purchase order to fulfillment maps to how enterprise customers already buy branded merchandise. For a platform, that means the value shows up in two places at once: the HR buyer gets an experience that works, and the procurement buyer gets a sourcing process with quotes, approvals, and records instead of a pile of vendor invoices.
Integration readiness
A staircase, not a build.
Four steps in order. Each one earns the next. No deep integration work until a workflow has proved itself in a live pilot.
Integration is where partnerships go quiet. Scope grows, security review stalls, and a year later there is a connector nobody uses. We stage deliberately so that every step is small enough to say yes to.
Step one. CSV and secure export.
A scheduled export of a minimal field set is enough to run a complete program end to end. No API work, no security exception, no roadmap slot. Most pilots never need more than this.
Step two. Single sign-on.
Administrators and approvers reach GiftWorks through the customer's existing identity provider. Access follows the same joiner, mover, leaver discipline as everything else the customer runs.
Step three. API and webhooks.
Event subscriptions replace the scheduled file once the workflow is validated and the field set is settled. Near real time triggers, status write-back where the customer wants it, retries and reconciliation built in.
Step four. Marketplace.
Listing is the last step, not the first. We would rather earn it with shared-customer evidence, a completed security packet, and a responsible-AI review than ask for distribution before there is anything to distribute.
Pilot first, always
The pilot shape we propose is deliberately narrow.
- 60 to 90 days, CSV first, no engineering commitment from the platform.
- Two event types only: onboarding kits and work anniversaries. These are the highest volume, most repeatable, and most visibly broken today.
- Weekly metrics, shared openly: on-time delivery rate, approval cycle time, budget adherence against policy tiers, employee choice completion, cost per moment, and exceptions by cause.
- A written readout at the end covering what worked, what did not, and what the next step should cost.
We would rather show you a measurement plan and run an honest pilot against it than show you a page of logos.
Responsible AI alignment
Governance is the design, not the paperwork.
We are combining employee data with spend. Both of those deserve gates, limits, and a log.
Agentic software that touches employee records and company money has to be built for review from the first day. Ours is. Here is what we would walk a responsible-AI reviewer through.
Bounded agents.
Six agents, each with one job, defined inputs, and defined authority. Lifecycle Listener builds the queue. Budget and Policy applies tiers and flags exceptions. Personalization curates choices. RFQ/Procurement runs the quote workflow. Fulfillment coordinates production and delivery. Insights reports outcomes. None of them can act outside its own scope, and none can widen its own permissions.
Named human gates.
Every consequential step stops at a person. HR approves which event types trigger anything at all. HR and finance own the budget policy, and exceptions always escalate rather than resolve themselves. The employee or the manager makes the final selection. Procurement approves the vendor and the quote. Operations owns escalations. Leadership reviews the program.
Explainability.
Every recommendation carries its reason: which event triggered it, which policy applied, which tier and cost center it drew from, which suppliers quoted, and why one was chosen. An approver sees the basis for the decision on the same screen as the decision.
Append-only logs.
Every action is written once and never rewritten. Trigger, policy evaluation, approval, selection, order, shipment, exception. The log is the audit trail, and it is exportable.
Data minimization.
The workflow needs event type, effective date, name, work location, department or cost center, and eligibility flags. It does not need compensation, Social Security numbers, or performance detail, so we do not request them. Least privilege by role, retention limits by field, and per-employee pause and suppress controls for anyone who does not want to be included.
Bias mitigation posture.
Eligibility and budget are determined by policy the customer writes, not by a model's judgment about who deserves recognition. Personalization affects the selection of merchandise, never whether a person qualifies. We monitor program distribution across department, location, and tenure so that a customer can see whether a policy is landing evenly, and we treat an uneven pattern as a finding to raise rather than a metric to optimize quietly.
We publish this before anyone asks. The privacy question is the first one a serious partner raises, and answering it early is the fastest way to become credible.
The governance memo and autonomy model are available on request for review teams.
Talk to our partnerships team.
The fastest way to judge this is to walk one lifecycle event end to end with us.
A first conversation takes about thirty minutes. We bring the integration staircase, the pilot workflow, the metrics we report against, and the human gate model, and we walk them against the event types your customers actually run today.
If the direction holds up, the next step is a narrow pilot. CSV first, 60 to 90 days, two event types, weekly metrics, and a written readout at the end. Small enough to run inside a quarter, specific enough to judge on the record it produces.