How Stitch Fix uses AI: algorithms plus human stylists
Stitch Fix pairs recommendation algorithms with about 1,700 part-time stylists, and has added generative AI tools such as Vision. How the hybrid model works and what the numbers show.
KEY TAKEAWAYS Summary by the editors
- Stitch Fix ranks inventory for each client with algorithms and then routes the shipment to a human stylist who makes the final selection, a hybrid model the company calls central to its business.
- As of 2 August 2025 Stitch Fix employed approximately 1,710 stylists, all part-time and paid hourly since January 2024, out of 4,165 employees.
- In autumn 2025 Stitch Fix introduced an AI Style Assistant, Stitch Fix Vision (generative AI outfit images based on client photos) and Stylist Connect messaging.
- For fiscal 2026 Stitch Fix reported net revenue of 1,348.1 million dollars, up 6.4 percent, with 2.277 million active clients (down 1.4 percent) and revenue per active client of 592 dollars (up 7.8 percent).
- Stitch Fix said in September 2026 that Vision had generated more than 22 million images for clients, but it has not isolated the revenue effect of its AI tools.
Stitch Fix uses AI to decide which clothes to offer each client and then lets a human stylist make the final choice. Its recommendation algorithms rank inventory using client-provided data and feedback, while about 1,700 part-time stylists curate each shipment, a hybrid model the company has run for years and has recently extended with generative AI tools such as Stitch Fix Vision and an AI Style Assistant.
What has Stitch Fix built?
Stitch Fix is an online personal styling service. Clients receive curated shipments called Fixes and can also buy directly through its Freestyle shop. In its fiscal 2025 annual report the company states that "AI and data science are integrated across almost every facet of our business." Its public algorithms explainer describes systems for:
- Styling: ranking items for each client before a stylist chooses.
- Stylist matching: assigning clients to stylists based on history and style affinity, with custom-built interfaces that help stylists understand each client.
- Demand modelling: anticipating future demand, since inventory is bought months ahead.
- Inventory management: balancing ordering and carrying costs and allocating stock across warehouses.
- Hybrid design: recombining attributes of existing styles, with client feedback as the measure of success, before designers refine the result.
Since autumn 2025 the company has added three client-facing tools: an AI Style Assistant offering conversational outfit ideas, Stylist Connect for two-way messaging with stylists, and Vision, which uses generative AI to show personalised outfits on an image based on the client's own photos.
Why did Stitch Fix invest in AI?
The business model depends on it. Sending a box of items the client did not choose only works if most of them fit and appeal, otherwise returns and shipping costs destroy the margin. Algorithms make it possible to narrow a large inventory to a short list per client at scale, while stylists add judgement that is harder to automate.
The more recent generative AI investments came during a period of shrinking client numbers. Fiscal 2025 revenue fell 5.3 percent, and CEO Matt Baer has framed the new tools as part of delivering "ultra-personalization at scale" and a more client-centric experience. For fiscal 2027 the company has said its outlook reflects investments in advertising and technology, including AI.
How does it work (data, models, process)?
Stitch Fix emphasises that "the vast majority of our client data is provided directly and explicitly by the client, rather than inferred, scraped, or obtained from other sources". Clients state sizes, budgets and preferences and give feedback on every item they keep or return. Merchandise is encoded with basic attributes such as brand, size, colour, pattern, silhouette and material, plus more nuanced descriptors.
According to the algorithms explainer, the styling system combines collaborative filtering, mixed-effects models, neural networks for visual similarity and natural language processing to rank inventory. "Once this machine ranking is complete, the shipment request gets routed to a human," who makes the final selection. In September 2026 the company said it also uses generative AI to support stylist recommendations, labour planning and customer service chats.
| Year | Milestone | Source |
|---|---|---|
| January 2024 | Move to a part-time only stylist model, paid hourly | Stitch Fix Form 10-K fiscal 2025 |
| 2 August 2025 | Fiscal 2025 ends with about 2.309 million active clients and about 1,710 stylists; revenue down 5.3% | Stitch Fix Form 10-K fiscal 2025 |
| Autumn 2025 | AI Style Assistant, Stylist Connect and Vision (beta) introduced | Digital Commerce 360 |
| 23 September 2026 | Fiscal 2026 revenue 1,348.1 million dollars, up 6.4%; Vision has generated more than 22 million images | Stitch Fix results release; MarketBeat |
What results has Stitch Fix reported?
As a listed company, Stitch Fix reports more numbers than most fashion retailers, but they describe the whole business rather than AI in isolation.
| Metric | Fiscal 2026 | Change year on year |
|---|---|---|
| Net revenue | 1,348.1 million dollars | +6.4% |
| Active clients (year end) | 2.277 million | -1.4% |
| Net revenue per active client | 592 dollars | +7.8% |
| Q4 net revenue | 324.4 million dollars | +4.2% |
On AI specifically, the company said in September 2026 that Vision had generated more than 22 million images for clients, that adoption had exceeded expectations and that clients who interact with it continue to show elevated Freestyle spending. It also said the share of customer chats resolved without additional support had more than doubled. Stitch Fix has not disclosed how much revenue or margin is attributable to these tools, and the elevated spending among Vision users may partly reflect which clients choose to use it.
What are the limits and open questions?
The hybrid model has costs. Stylists remain a large workforce, and the shift to part-time hourly contracts in 2024 shows the pressure to make human curation cheaper. Generative images of clients wearing outfits raise questions about how accurately fit is represented and how personal photos are stored and used. And growth in revenue per client has so far coincided with a slightly smaller client base, so the AI tools have not yet been shown to reverse client losses on their own.
Analysis: generative AI also blurs the division of labour that made the model distinctive. If software now supports stylist recommendations, writes chat replies and visualises outfits, the company has to keep showing clients what the human stylist adds, or the service risks becoming an algorithmic shop with a premium price.
What can other fashion companies learn?
Stitch Fix is one of the few fashion businesses that has described its algorithms publicly for years, which makes it a useful reference for any brand or retailer that offers advice alongside products:
- Ask customers directly. Explicit, first-party preference and feedback data is more reliable than inferred signals and easier to justify to clients.
- Let machines rank and people decide. Using algorithms to shortlist and humans to choose is a practical pattern for any curated or advisory sale.
- Connect personalisation to buying. Stitch Fix links recommendations to demand modelling and inventory, so personalisation informs what is bought.
- Measure carefully. Even with detailed reporting, separating the effect of AI from other changes is hard, so controlled tests matter.
Frequently asked questions
Does Stitch Fix use AI or human stylists?
Both. Algorithms rank inventory for each client, and a human stylist then makes the final selection and personalises the shipment. As of August 2025 the company employed about 1,710 stylists, all part-time.
What is Stitch Fix Vision?
Vision is a generative AI feature launched in beta in autumn 2025. Clients upload photos, and the tool creates images of them in personalised, shoppable outfits. By September 2026 it had generated more than 22 million images, according to the company.
What data does Stitch Fix use for recommendations?
Mainly data clients provide directly, such as sizes, budgets, style preferences and feedback on items they keep or return. This is combined with detailed product attributes like colour, pattern, silhouette and material.
Is AI helping Stitch Fix grow?
Fiscal 2026 revenue rose 6.4 percent and revenue per active client rose 7.8 percent, while active clients fell 1.4 percent. The company links its AI tools to higher engagement and spending, but it has not disclosed how much of the growth they explain.
One edition every weekday morning. Read in five minutes. Free for industry professionals.
