Korea Kolmar has launched Kolmarpkg.com, a global packaging sourcing platform that uses AI to analyse container design, materials and functionality, and to match a brand’s requirements against thousands of pre-verified options. The company’s headline claim is that it compresses packaging selection from a process that routinely runs several months into one that can close inside a month.
For a manufacturer to build this rather than a packaging supplier is the detail worth noting. Kolmar sits on the ODM side, which means it has watched the same sourcing bottleneck delay client launches for years and has a direct commercial interest in removing it.
What the platform does
The core function is search against a verified catalogue. A brand can filter by package type, capacity and country of origin, and can run image-based searches — upload a reference container, get structurally similar options back. From there the workflow continues into supplier connection and cost estimation, and for clients who want it, into Kolmar’s own ODM service.
Sustainable materials are given prominence in the catalogue, which reflects where buyer demand has moved rather than any technical property of the search system. Recyclable and mono-material containers are what brands are being asked for by retailers, and a sourcing tool that surfaces them first is meeting a real filter requirement.
Why sourcing takes months in the first place
The six-month figure sounds implausible until you break the process down. Very little of that time is spent searching. It’s spent on the steps around the search.
A brand typically starts with a reference — a container they saw, a mood board, a competitor’s product. Finding out who makes something similar means working through trading companies, agents and personal contacts, because the vast majority of packaging vendors have no meaningful digital catalogue. Once candidates exist, samples ship physically, and international sample shipping alone eats two to four weeks per round. Most projects need two or three rounds.
Then come the questions that actually kill timelines: is this container compatible with the formula, will the pump handle that viscosity, what’s the minimum order quantity, does the decoration method survive the fill line, and is the mould tooling existing or new. Each of those is a separate conversation with a separate party, and they mostly happen sequentially rather than in parallel.
A search platform attacks the first block of that chain — discovery and shortlisting — very effectively. It does nothing about sample shipping or compatibility testing, and it shouldn’t be expected to.
Where the compression is real
Reading the claim carefully: the platform can plausibly take a discovery-and-shortlist phase that ran eight to twelve weeks down to days, because the catalogue is pre-verified and the filters do work a human agent would otherwise do by memory and phone call. That’s a genuine and substantial gain.
What it cannot compress is the physical and the relational. Samples still need to be held. Fill-line trials still need to run. MOQs still need to be negotiated with a supplier who has their own capacity constraints. A brand that reads “one month” as end-to-end and plans a launch calendar around it will be disappointed at the point where a pump fails a viscosity test and the whole shortlist reopens.
The honest framing is that this removes the least valuable weeks from the process, which is exactly the right weeks to remove.
The open questions
Three issues are being raised by people in the industry, and none of them are resolved:
- Recommendation quality. An image-similarity match returns containers that look alike. Looking alike and behaving alike under a specific formula are unrelated properties, and a system that conflates them will confidently propose unsuitable options.
- Design and copyright exposure. Uploading a reference image of a competitor’s container, and receiving close matches, sits in uncomfortable territory. Packaging design carries real IP in several markets, and a tool that makes visual imitation frictionless will eventually produce a dispute.
- Who this actually helps. Larger brands already have sourcing teams and supplier relationships; the marginal gain for them is modest. Smaller brands stand to gain far more, but they’re also the ones least equipped to evaluate whether a recommended container is right, and most likely to be blocked by minimum order quantities the search interface doesn’t surface.
Data handling is a fourth question, less discussed but more consequential over time. A platform that sees which containers a brand is shortlisting is holding early-stage product intelligence. That’s commercially sensitive in a way a packaging catalogue never was.
The rest of the AI cosmetic packaging picture
Sourcing is one application among several, and they’re at very different stages of maturity.
Retailer-led recycling programmes now use automated sorting to separate returned containers by material — Sephora’s take-back initiative is the most visible example. That’s operational and working, and it’s the least glamorous of the applications.
Personalised packaging is further out. The idea is that a brand generates container designs or decoration tailored to an individual customer, often connected to a skin diagnostic that also drives the formula recommendation. Digital printing makes short runs economically possible; what remains expensive is the fulfilment complexity of a genuinely one-off SKU.
Smart packaging — containers reporting usage, remaining volume or storage temperature back to an app — has been demonstrated repeatedly and adopted rarely. The unit economics don’t work at mass-market price points, and they may not for a while.
Production-side AI is the quietest and the most established. Vision-based defect detection on fill and decoration lines is standard equipment in serious facilities, and it reduces reject rates without anyone writing a press release about it.
Market context
The cosmetic packaging market is commonly estimated at around $1.63 billion in 2023 with forecasts reaching roughly $2.5 billion by 2032, against a broader beauty market growing in the region of 5–7% annually. Skincare remains the largest slice, with figures around $180 billion cited for the sector.
Those numbers explain the investment logic. Packaging is a small market attached to a very large one, and it’s the part of the large one where lead times are worst. Any tool that shortens time-to-shelf for a K-beauty product heading into an export market is attacking a bottleneck with disproportionate value.
How to use this if you’re sourcing
Use a platform like this to widen the shortlist, not to shorten the evaluation. The value is in seeing forty viable containers instead of the six an agent happened to remember. Then evaluate those forty with the same rigour you’d have applied to the six.
Ask about minimum order quantity before you fall in love with a container. It’s the constraint that most often eliminates an otherwise perfect option, and it’s rarely visible in a catalogue listing.
Send the formula to the packaging supplier, not just the specification sheet. Compatibility problems — corrosion, permeation, pump failure with high-viscosity or high-oil formulas — surface in physical testing and nowhere else.
And keep the sourcing relationship even if the discovery moves online. When a mould breaks or a shipment is short three weeks before launch, the thing that resolves it is a supplier who knows your name.
“Sourcing timelines are one of the most common pain points brand partners bring to us, so a tool that promises to compress packaging selection from months to weeks is the kind of development our clients notice immediately. In practice, we tell brands not to expect AI matching to replace the relationship-building part of sourcing — supplier trust, minimum order quantities, and real sample turnaround still matter as much as they ever did. What’s genuinely useful here from a marketing lens is the transparency layer: being able to show a buyer a wider, well-organized set of packaging options earlier in the conversation shortens the back-and-forth considerably, even if the final decision still comes down to touch, feel, and cost.” — Hyejin Cho, Digital Marketing Specialist, Marketing and Communication Team at SPSCOS