I asked four AI models which notification infrastructure to use in 2026. Three of them still send you to Twilio.

We build Bersyn, a tool that tracks which products AI models name when someone asks for a recommendation in a category. This week's scan was notification infrastructure, the API layer you reach for when your app needs to send email, SMS, push, in-app and Slack messages without building all of it yourself. The result was familiar in a way that should worry anyone building in this space.

The question we tested is the one a developer actually types: "what is the best multi-channel notification API for a B2B SaaS platform." We asked it five ways across ChatGPT, Claude, Gemini and Perplexity, then counted how often each modern tool got named.

            ChatGPT   Claude   Perplexity   Gemini

Knock          0%       80%        80%        20%

Novu          20%       80%        40%        80%

Courier        0%       60%       100%        20%

Read the ChatGPT column. Knock: zero. Courier: zero. Novu: barely. These are the tools built for exactly this job, and on the model most people open first, they are not in the answer at all.

Who gets named instead? Twilio and OneSignal, almost every time. Here is ChatGPT, verbatim, on the best multi-channel notification API:

Choosing the best multi-channel notification API for a B2B SaaS platform depends on several factors... Here are some popular options: 1. Twilio: Known for its robust SMS and voice capabilities...

Twilio is a fine company. But it is the SMS-and-voice incumbent, not the modern notification-infrastructure layer, and a developer reading that answer walks away never knowing Knock, Novu or Courier exist.

Claude is the outlier again

One model broke the pattern. Claude named Knock in 80% of its answers, Novu in 80%, Courier in 60%, and it literally opens with "Best Multi-Channel Notification APIs, 1. Knock, best for modern B2B SaaS teams." It knows the category as it actually exists today. The other three are running on an older map.

If you read my deploy teardown this will feel familiar, because it is the same shape. Ask about hosting and ChatGPT says Heroku. Ask about notifications and it says Twilio. The incumbent that dominated the blog posts and Stack Overflow answers a few years ago is the one the training-data models echo, and the tool you actually build on now is invisible.

Why this matters if you build dev tools

The reflex is to wait it out. The models will catch up. Maybe. But your buyer is asking today, and today the answer routes them to the incumbent. Whether your product even exists to that buyer comes down to which model they happened to open, and three of the four send them elsewhere.

This is not a competitor outranking you on features. It is the AI quietly leaving you out of the shortlist, with no signal that it happened. You do not lose the deal, you never enter it.

What changes it

We do not claim AI is biased or that anyone pays for placement. We read what the models say and show the receipts. What moves the answer over time is the same boring thing that moved search: specific, verifiable, third-party content that ties your company to the category and the exact question buyers ask. Comparison pages. Honest docs. Other people naming you in the words a buyer would use.

The first step is not publishing more. It is finding out what the models say about you right now, so you know whether you are omitted, described generically, or named only by the one model nobody opens.

See your own category

We built Bersyn to show you exactly the table above, for your company, with the verbatim answers behind it. Run a free scan at bersyn.com and see which models name you, which name a competitor in your slot, and why.

If you ship on Knock, Novu, Courier or anything in this space, tell me in the comments which model gets your category right. The disagreement between them is the most interesting thing I look at.