How lululemon approaches AI in product, planning and inventory
lululemon has named a chief AI and technology officer and set a new-style target, but has published few details on AI in planning. Here is what is on record.
KEY TAKEAWAYS Summary by the editors
- In September 2025 lululemon said it had appointed Raju Das as its first chief AI and technology officer, with a mandate to use AI and technology to speed up its product innovation process.
- lululemon said it planned to raise the share of new styles in its assortment from 23 percent to 35 percent by the following spring, which is the clearest public target linked to its AI and technology agenda.
- lululemon's annual report for the year ended 1 February 2026 lists forecasting guest demand, data use and AI adoption as risks, but does not describe specific AI planning or personalisation systems.
- On 7 October 2026 the company said the role of chief technology officer, covering product and supply chain technology, cybersecurity and AI use, was open as part of a leadership restructure under CEO Heidi O'Neill.
- No public source reviewed here quantifies results from AI at lululemon, so claims about AI-driven demand planning or personalisation should be treated as unverified.
lululemon has publicly tied AI to a business problem, getting new products to market faster and in the right quantities, but it has disclosed little about the systems involved. The record shows a senior AI and technology appointment in 2025, a target to increase the share of new styles, and annual report language that treats AI as both an opportunity and a risk. It does not show published results from AI-based demand planning, inventory allocation or personalisation.
This case study sets out what lululemon has said, what its filings add, and what remains undisclosed, so that readers can separate documented facts from the wider commentary on the company.
What has lululemon said publicly about AI?
In its second quarter earnings call on 4 September 2025, lululemon described a new role, chief AI and technology officer, filled by Raju Das, who started on 2 September and reports to the CEO then in post, Calvin McDonald. According to PYMNTS, McDonald called it a role with an elevated mandate to enable AI and technology to help expedite the product innovation process.
The context was commercial rather than technical. McDonald said the go-to-market process sometimes failed to meet demand for new styles, and attributed part of the weakness in the Americas to product life cycles that had run too long and become too predictable. In the quarter ended 3 August 2025, Americas comparable sales fell 4 percent, international rose 15 percent and total comparable sales rose 1 percent.
The article names no AI tools, vendors, budgets or outcomes. The public statement is therefore one of intent: use AI and technology to shorten design and go-to-market cycles.
How does AI connect to lululemon's new-style target?
lululemon said it aimed to lift new styles from 23 percent to 35 percent of its assortment by the following spring. A higher share of new styles raises the planning problem: there is less sales history for each item, so forecasts lean more on analogues, early sell-through signals and judgement. This is where AI-assisted forecasting is usually proposed, but lululemon has not said it is using it for that purpose.
The annual report for the year ended 1 February 2026 reinforces the link between product newness and risk. It describes increasing new style penetration through product innovation as part of its Americas strategy, and warns that failing to forecast guest demand accurately can leave excess inventory that may need to be written down.
| Area | What the public record shows | Status |
|---|---|---|
| Governance | First chief AI and technology officer announced August 2025; CTO role open in October 2026 | Documented |
| Product development | Stated aim to speed up design and go-to-market; target of 35 percent new styles | Intent documented, method not disclosed |
| Demand planning | Forecasting risk disclosed; no AI forecasting system described | Not disclosed |
| Inventory | Excess inventory risk disclosed; no AI allocation system described | Not disclosed |
| Personalisation | No description of AI personalisation in the sources reviewed | Not disclosed |
| Risk and regulation | AI security risks, EU AI Act and data sovereignty cited in annual report | Documented |
What does the annual report say about AI risk?
The filing for the year ended 1 February 2026 treats AI mainly as a risk and capability question. It states that the company may not efficiently and effectively implement and leverage technological advancements such as AI, and that its increasing use of AI tools, including generative AI, introduces new security risks.
It also notes that emerging laws, including the European Union AI Act, may apply, that sovereign AI initiatives could limit the ability to deploy centralised AI tools globally, and that third-party AI platforms increasingly influence product discovery and purchases on behalf of customers. That last point matters for any fashion brand: shopping assistants outside the brand's control may become a channel in their own right.
Separately, the company warns that it may not have or successfully leverage relevant data to understand consumer preferences. That is a data-readiness warning, and it is consistent with the general finding that AI projects depend on clean, connected product and sales data.
What changed in leadership in October 2026?
Retail Dive reported on 7 October 2026 that CEO Heidi O'Neill, named in April 2026 and in post since 8 September, had restructured the senior team. Maggie Gauger becomes chief product officer, overseeing merchandising, design, footwear, product innovation and materials science. Joseph Godsey becomes chief operating officer, with sourcing, production and fulfilment, and a stated focus on product quality and speed to market. The chief supply chain officer is leaving.
Several roles are open, including chief technology officer, who is to oversee product and supply chain technology, including cybersecurity and AI use. The article gives no inventory figures and no AI project details. What it does show is that the open technology role is described alongside product and supply chain responsibilities in a broader reorganisation.
What can other fashion companies learn from this?
- Start from the commercial problem. lululemon framed AI around product newness and speed to market, not around a technology list.
- Expect leadership churn to affect AI programmes. A new CEO and an open technology role show that ownership can shift within a year.
- Treat claims with care. Third-party articles describing lululemon's AI personalisation or planning often cite no company source. Prefer filings and earnings calls.
- Plan for risk alongside value. Security, regulation and data sovereignty appear in the filing next to the opportunity.
What remains undisclosed. Public sources do not say which planning or merchandising systems lululemon uses, whether any forecasts are produced by machine learning, how AI affects allocation by region, or how guest data is used for personalisation. Nor do they report measurable gains in forecast accuracy, markdown rates or speed to market.
How should readers judge claims about lululemon and AI?
Because the company has said little about methods, claims in the wider press fall into three groups. The first is what lululemon itself has said, such as the appointment, the target for new styles and the mandate to expedite product innovation. The second is what filings say, which is risk language and strategy. The third is inference by commentators about what a retailer of lululemon's size must be doing. Only the first two are evidence.
A practical test is to ask whether a statement names a system, a process and a measured result. A statement such as AI improves our forecasts names none of these. A statement such as forecast error for new styles fell by a stated amount after a defined change would name all three. In lululemon's case, none of the sources reviewed reaches that standard.
What would AI in apparel demand planning have to solve for a brand like this?
A brand that wants a larger share of new styles faces a specific planning problem. New items have no sales history, so forecasts must be built from similar past items, early test results and merchandising judgement. Sizes and colours multiply the number of lines to plan. Assortments with shorter lives leave less time to correct a wrong buy through reorders.
- Data needed: clean item attributes, historical sell-through by size, colour and region, and consistent records of promotions and stock-outs.
- Decisions supported: initial buy quantities, regional allocation, and when to chase or cut a style.
- Main limits: sparse data for genuinely new designs, shifts in taste that history cannot show, and the need for planners to explain and override outputs.
- Costs: data engineering, change management and ongoing monitoring, which often exceed the cost of the model itself.
None of this is specific to lululemon. It describes the conditions under which any apparel brand would test AI planning, and it is useful as a checklist when reading the company's future disclosures.
Frequently asked questions
Does lululemon use AI for demand planning?
lululemon has not published a description of AI demand planning. Its filings discuss the difficulty of forecasting guest demand and the risk of excess inventory, and it has appointed a senior AI and technology leader, but no forecasting system or result is documented.
Who is lululemon's chief AI and technology officer?
In August 2025 lululemon announced Raju Das as its first chief AI and technology officer, starting in September 2025. In October 2026 the company listed the chief technology officer role as open in a leadership restructure, so readers should check the current organisation chart.
What is lululemon's new-style target?
In September 2025 the company said it planned to increase new styles from 23 percent to 35 percent of its assortment by the following spring. It linked this to faster design and go-to-market processes.
Does lululemon say AI affects its inventory?
Not directly. Its annual report warns that inaccurate demand forecasts can cause excess inventory and write-downs, and that it must leverage data well, but it does not attribute inventory outcomes to AI.
One edition every weekday morning. Read in five minutes. Free for industry professionals.
