The $29 AI writer is gone

Search for Hypotenuse AI and most of what comes back describes a cheap blog-and-copy tool: a $29 entry plan, 50,000 words, a plagiarism checker, a Jasper alternative for people who did not want to pay Jasper prices. That product no longer exists. Checked on October 23, 2026, the pricing page lists two plans, Basic and Ecommerce Enterprise, and both are priced "Custom". There is no dollar figure anywhere on the page and no monthly or annual toggle.

What replaced it is a pitch to a different buyer. Hypotenuse is a Y Combinator Summer 2020 company that YC now lists as an "AI-native Operating System for Enterprise Ecommerce Companies", and it sells itself as an "AI-first PIM": software that fills in missing product attributes, tags and categorizes SKUs, rewrites product descriptions in bulk, cleans up product photos and pushes the result to Shopify, Akeneo, Salsify or Salesforce Commerce Cloud. Blog articles survive as one of "20+ other ecommerce content types". They are no longer the point.

The question now is narrower: does a demo pay off for a catalog of thousands of SKUs, and where should a writer go instead? Product images below are Hypotenuse's own, from hypotenuse.ai.

Pricing: what you can see and what you cannot

Basic is described as "for ecommerce teams getting started with AI product content". It comes with one seat, 40+ languages, the product description generator, the other content types, a limited allowance of HypoChat (the built-in chat assistant with web research) with 30 document uploads, and 1,000 images on the premium HypoArt text-to-image model. The word allowance is not stated. The plan card ends with "Contact us for more seats or words", which tells you a cap exists without telling you where it sits.

Ecommerce Enterprise is the real product. It adds attribute enrichment, taxonomy management, tagging and categorization, an AI quality checker for content and product data, a bulk image editor, bespoke AI models, custom PIM and ERP integrations, team access controls and a dedicated account manager. The FAQ says there is "no fixed cap" on SKUs and that customers manage "anywhere from a few thousand to millions of SKUs". The price is whatever the demo call settles on.

For reference, the archived pricing page from January 12, 2026 still showed the self-serve Marketing & SEO plans: Entry at $29/month ($19 billed annually) for 50,000 words, Essential at $87/month ($56 annually) for 250,000 words, and Blog Pro from $230/month ($150 annually) with three seats. Between January and October 2026 all of those disappeared, along with the five-seat Ecommerce Pro tier, and we found no announcement of the change. If a review gives you a dollar figure for Hypotenuse, it is out of date.

The trial and the refund policy

The January page promised "No credit card required". The current page says only "Try for free", and the terms of service (last updated June 24, 2025) say billing information may be required and that on the last day of a trial, "unless you cancelled your Subscription, you will be automatically charged." Hypotenuse does not state the trial length anywhere we could find. There is no money-back guarantee either. The only refund wording in the terms is "no refunds will be issued", and it applies to accounts blocked for unreasonably high usage or shared logins. Set a calendar reminder before you start the trial.

What it does to a product catalog

The core loop is enrichment. You bring products in by CSV or XLSX or through a connector, and Hypotenuse fills empty attributes from what it calls references: "image, webpage scraping, UPC, or factsheets." In its own demo, a row of bike SKUs gets Frame, Brakes and Tyres values filled from those sources. The same record then feeds tagging, categorization into your taxonomy and, last, the description, written in a brand voice built from your guidelines.

Bulk work is where it earns an enterprise price. Bulk generation writes or rewrites descriptions for a selected set of products in one pass, and the newer AI agents chain the steps: they "clean up your product data, enhance images, and craft product descriptions in your brand's voice", with a human review step and automatic flags for low-quality images, mismatched attributes or underperforming descriptions. For a merchandising team that review queue matters more than the prose. A wrong attribute in a filter facet hides a product from shoppers. A flat sentence only bores them.

The image tools cover background replacement, upscaling, cropping, in-context product shots and a compliance checker. They will not replace a campaign photographer. They are aimed at 4,000 supplier photos with mismatched white backgrounds.

Integrations and API

The integrations page lists Akeneo, Mirakl, NetSuite, Plytix, Salesforce Commerce Cloud, Salsify, Shopify and Stibo on the commerce side, and WordPress and Webflow for content. Shopify gets one-click publishing of generated descriptions. BigCommerce and WooCommerce are not on the list, although several older reviews name them; on those platforms you are working through CSV round-trips or the API. The API is pitched as a way to "send product information and retrieve back product descriptions for your PDP". You get a key by request, and there is no published API pricing.

Your data, the model and the fine print

The data promise is worded carefully. The pricing FAQ says your data is "never shared with third parties or used to train public AI models", and adds SOC 2 Type II compliance and encryption in transit and at rest. "Public" is the word to notice. Enterprise sells bespoke AI models, which are presumably tuned on the customer's own catalog. Ask in the demo where that training data goes.

Output rights are clean: the terms say you "retain any and all of your rights to any Content you submit, generate, post or display." Before signing, note that subscribers grant Hypotenuse a revocable licence to show their name and logo in its marketing, and the contracting party is Hypotenuse Technologies Pte. Ltd. under California law. The privacy policy keeps personal data "only for as long as is necessary", with no fixed retention period, and deletion requests go to support by email.

None of the pages we checked name the language model that does the writing, or say whether it changes over time. For a team that signs off brand copy, a silent model swap means a silent shift in tone across thousands of product pages. Get that answer in writing.

What users and customers report

Independent feedback is thin. G2 shows around 4.7 out of 5 from roughly 73 reviews. The praise that recurs is time saved and natural-sounding copy. The complaints are that "the price is a bit too high on a monthly basis", that longer output repeats sentences, and that the free trial has too few credits to judge it properly. Trustpilot had no reviews at all when we looked on October 23, 2026, and the old help centre at help.hypotenuse.ai now redirects to the homepage, so documentation sits behind the sales process as well.

The customer stories page carries the big numbers. Crunchyroll Store reports "3x output and a 46% YoY increase in conversion", Volcom "50% cost and 83% time savings", and Australian Defence Apparel says it cut time to market by over 95%. These are Hypotenuse's own case studies, not audited results. The names are telling all the same: apparel and merchandise retailers whose catalogs turn over every season.

Verdict: who should book the demo, and who should skip it

Book the demo if you manage thousands of SKUs on Shopify, Akeneo, Salsify, Salesforce Commerce Cloud or a similar stack, your product data has gaps that hurt filtering and on-site search, and descriptions are a backlog rather than a creative task. Enrichment from photos and spec sheets, a review queue that flags bad attributes, and bulk image editing in one tool are a real reason to pay enterprise money. Ask for a pilot on 500 of your own products and judge it on attribute accuracy, not on how the copy reads.

Skip it if you are a blogger, a content marketer or a small store. You would be paying a price you cannot see in advance, on a trial that may charge your card automatically, for a tool whose blog writer is now a side feature. For long-form SEO writing, Jasper and Writesonic publish their prices and are built for that job. A Shopify store with a few hundred products is better served by Shopify Magic, which comes with Shopify plans, or by a per-product tool such as Describely, which third-party comparisons list at $0.75 per product for the first 500 (check its own pricing page before buying).

Hypotenuse made a clear bet: stop competing with every AI writer on price and sell product-data cleanup to retailers who will pay for it. For that buyer an hour on a demo call is well spent. For anyone who found it in an old "best cheap AI writer" list, it is no longer that product.

Try Hypotenuse AI