Why a Self-Hosted AI Radio Should Rearrange How You Think About Product Discovery
I’ve been watching e-commerce tooling long enough to notice when a product that has nothing to do with selling suddenly exposes a pattern that every DTC operator should steal. SUB/WAVE is that product. It’s a self-hosted radio station that turns a personal music library into a 24⁄7 broadcast with an AI DJ, live weather reads, song requests, and—critically—no skip button. For a cross-border seller, the surface-level reaction is “cute, but I run an Amazon store, not a college radio station.” That reaction would miss the point. The real insight is about how AI can transform a static catalog into a living, shared experience that forces attention and discovery. Most sellers are sitting on thousands of SKUs that function like an unplayed music library: full of potential, but the algorithmic recommendation engines only surface the same top sellers. SUB/WAVE’s philosophy—communal listening, AI-driven curation that respects inventory depth, and full local control—offers a blueprint for breaking out of the personalization trap and building something closer to brand community. If you’ve ever wondered why your repeat purchase rate is flat despite endless A/B tests on product recommendation widgets, read on.
What Problem SUB/WAVE Actually Solves (and Why It’s Your Problem Too)
The core problem SUB/WAVE addresses is the loneliness of infinite choice. Streaming apps gave us every song ever recorded and replaced radio’s shared moment with per-user algorithmic bubbles. As maker Parminder Klair put it, “Streaming apps gave us infinite choice and somehow made music lonely.” That same loneliness infects e-commerce. Every customer lands on your product page, gets a personalized recommendation block, and is stuck in their own private discovery funnel. You optimize for conversion on that single page, but you never create a reason for anyone to browse broadly, stay longer, or feel like they’re part of something larger than a transaction.
SUB/WAVE solves that by enforcing a single broadcast: “No skip button, no per-listener shuffle. You tune in and hear whatever’s on.” That constraint is radical because it forces consumption of the deep catalog. The AI DJ doesn’t just play your top 40; it pulls from rarely played tracks, even “dig[s] something out of the back of the library and talk[s] about why it’s been sitting there” (see the product hunt comments where Parminder describes a “deep-cut skill”). For an e-commerce operator, that’s the equivalent of an AI host who, in the middle of a live stream, picks your worst-selling SKU from three years ago and convincingly explains why a customer needed it today.
The cross-border seller’s problem is identical: you have catalog bloat. You import 500 SKUs from a factory in Shenzhen, 80% never get a second look after launch. You spend ad dollars on the top 20% and let the rest rot in fulfillment centers. SUB/WAVE suggests an alternative: build a digital “radio station” for your product catalog—a curated, shared experience that cycles through everything you sell, with an AI host that can narrate fit, usage, and cross-sell opportunities in real time. The tooling to do this exists today: LLMs, TTS, and a content management system that treats each product like a track in a playlist.
How It Differs from the Incumbents (Spotify, Pandora, and the Amazon Recommendation Engine)
The obvious comparison is Spotify’s AI DJ, which launched in 2023. That product picks songs based on your listening history and throws in some canned voiceover about your taste. It’s personalized, opaque (Spotify owns the model), and runs entirely in the cloud. SUB/WAVE is the opposite: the AI DJ is open-source, runs locally via Ollama, uses local TTS, and the station is the same for everyone tuning in. As Parminder noted, “If you run Ollama and local TTS, nothing leaves the box at all. The only thing it phones out for is the weather.”
For an Amazon seller, this distinction matters more than for a Shopify operator. Amazon’s recommendation algorithm is a black box that pushes your highest-margin items whether you like it or not. You have zero control over what a shopper sees next. SUB/WAVE’s approach gives the “station owner” complete editorial control over the sequence, the transitions, and the narrative. That’s the difference between renting shelf space on Amazon and owning a broadcast channel. In a Shopify context, where you control the entire front-end, you could theoretically build a “product radio” using Shopify’s Storefront API combined with a local LLM stack—but few do because the mental model of “curated broadcast” hasn’t penetrated e-commerce yet.
SUB/WAVE also differs from traditional radio (which is passive linear content) by being dynamically generated in response to listener requests. The AI DJ accepts “plain-language requests” and adjusts in real time. That’s a level of interactivity that no product recommendation widget offers. Imagine a live shopping stream where viewers type “show me a waterproof backpack under $50” and an AI host instantly pulls the matching SKU, tells a story about its durability, and highlights a complementary item. That’s the future SUB/WAVE is pointing toward, and it’s closer than most sellers realize.
What Cross-Border Sellers Should Borrow from SUB/WAVE
1. Treat Your Catalog as a Library, Not a List
Most sellers organize their products into categories and hope search + ads fill the gaps. SUB/WAVE’s architecture explicitly surfaces “deep cuts” and “rarely-played tracks” as a way to rediscover dormant inventory. In e-commerce, that’s dead stock. Parminder described a picker that builds its candidate pool “from several sources, and deep cuts and rarely-played tracks are one of them.” You can replicate this logic in your product feed. Use a script to identify SKUs with fewer than 10 sales in 90 days, then cycle them into a “discovery showcase” on your homepage or in an email sequence. The goal isn’t to sell them all—it’s to change the browsing experience from “what do I already know?” to “what haven’t I seen?”
2. Use AI to Add a Human(ish) Voice to Product Descriptions
SUB/WAVE’s DJ doesn’t just play tracks; it does station idents, reads the weather, and banter with a co-host. The maker specifically mentions “DJ personas, guest co-hosts with banter, and produced programmes with per-episode plans.” If your brand sells across Amazon, TikTok Shop, and Shopify, the content you produce for each channel is usually a static product description. SUB/WAVE suggests adding an AI-generated audio or video host that walks through products in a narrative flow—especially in live-stream contexts where engagement depends on personality. The TikTok Shop algorithm rewards watch time, not just clicks. A scripted, AI-driven show that mimics radio pacing could keep viewers on the stream longer than a typical product demonstration.
3. Self-Host Your AI Stack for Data Sovereignty and Cost Control
SUB/WAVE is MIT-licensed and runs entirely locally. For sellers dealing with proprietary product specs, pricing strategies, and customer data, the idea of feeding everything into a cloud AI service like ChatGPT raises legitimate concerns. Leaked prompts, training data ingestion, and cost per API call add up. A self-hosted LLM via Ollama with a local TTS engine (like Piper or Coqui) can generate product descriptions, customer service scripts, and ad copy without sending a single byte to an external API. The upfront hardware cost (a decent GPU or a cloud VM with a T4) is a few hundred dollars a month—often less than what you’d pay for API calls from a single product catalog rewrite.
4. Design for Shared Experience, Not Personalization
The hardest pill for sellers to swallow is that personalization has diminishing returns. You optimize every page view into a unique snowflake, but you lose the serendipity of hearing a new song because someone else requested it. SUB/WAVE’s “one stream, everyone hears the same track” creates a sense of community that Amazon’s “customers who bought this also bought” can never replicate. You can test this by running a weekly “product radio” live stream on Twitch or TikTok where you play a fixed sequence of products—no skipping, no custom queues. The chat becomes the request line. I’d bet your average order value goes up because viewers stay longer and discover items they wouldn’t have found through search.
Where the Math Breaks (and What SUB/WAVE Gets Wrong for E-Commerce)
For all its charm, SUB/WAVE is not a product most sellers would implement as-is. The biggest limitation is the dependency on an existing, well-organized local library (Navidrome or any Subsonic server). As user Kamil noted in the comments, “most people now i guess is that we don’t store our music files, just use spotify or apple music.” Parminder didn’t hedge: “SUB/WAVE needs a library you actually hold, so if you’re all-in on Spotify or Apple Music it isn’t for you.” In e-commerce terms, that’s the equivalent of needing clean, structured product data with high-res images, formatted descriptions, and accurate inventory—something most sellers don’t have. If your catalog is a mess of dropshipped listings with supplier-sourced CSV files, you can’t point an AI at it and expect radio magic.
The “no skip button” also works for music but fails for commerce. Customers demand the ability to search, compare, and leave. Forcing them into a linear stream would create friction. The better application is optional: a “radio mode” toggle on product listing pages that plays through recommendations in a timed sequence, with the option to skip forward or exit. SUB/WAVE’s model of no escape is fine for background listening; it’s not fine for someone trying to buy a specific item.
There’s also the question of scale. Running a local LLM (even a small one like Llama 3.1 8B) requires a machine with at least 8 GB of VRAM. For a solo DTC operator, that’s doable. For a team of 50 with multiple stores across marketplaces, maintenance becomes a burden. The maker acknowledged 48 releases in three months—that’s impressive but indicates a fast-moving target. If you build a content pipeline that depends on this tool, you’re betting on its longevity and your ability to keep up with updates.
Why Amazon Sellers Should Care More Than Shopify Ones
Amazon sellers operate in a walled garden where you cannot change the checkout flow, the recommendation widgets, or the search ranking. A tool like SUB/WAVE is irrelevant if you can only touch your Amazon listings. However, you can use the concept off-platform: build an audience on YouTube or a private Discord server where you run a weekly “product radio” show that highlights your Amazon catalog. The AI DJ pre-rolls a description of the product, tells a story, and then directs viewers to the Amazon listing. That’s how you turn radio into a traffic source. Shopify sellers have the advantage of being able to embed the experience directly on their store. But Amazon sellers have more to gain from community building because they can’t rely on the platform’s internal discovery.
What I’d Watch / Test Next
If you’re a cross-border operator, do three things this week:
Run the live demo. Go to getsubwave.com/listen and spend fifteen minutes with it. Pay attention to how the AI DJ introduces tracks, transitions, and handles requests. Then imagine a version where each “track” is a product page, and the DJ’s voiceover is your brand’s tone of voice. Sketch the content flow for your top 20 SKUs.
Audit your dead inventory. Pull a list of SKUs with zero sales in the past 60 days. Manually pick five that have decent images and descriptions. Write a 30-second script for each, as if you were introducing them on a radio show. Record one with a simple TTS tool (like Edge’s built-in TTS) and upload it to TikTok or Instagram Reels. See if engagement differs from a standard product photo post. This is a fast, zero-cost validation of the “deep cut” concept.
Set up a local LLM for product content generation. Install Ollama on your machine or a cheap cloud VM. Pull a model like
llama3.1:8b. Feed it a CSV of your product titles and descriptions and ask it to generate a 60-second radio script for each, including a cross-sell suggestion. Compare the output to your current copy. The quality may surprise you—and you’ve learned a skill that will only become more valuable as AI tooling matures.
SUB/WAVE is a niche product for music hoarders. But the pattern it embodies—a shared, AI-narrated journey through a deep library—is directly applicable to the biggest unsolved problem in cross-border e-commerce: turning a flat catalog into a living brand experience. Don’t let the fact that it’s a radio station fool you. Listen carefully.
