Stop making chatbot happen

The recent announcement of Zalando fashion assistant looks like yet another chatbot. Let's take a look together at what tools and services you should build as a company using AI, and why chatbot is not it.


Mentioned Zalando's fashion assistant announcement is here.

ChatGPT has changed things

With the advent of general AI models (ChatGPT, Bing, Bard), and the emergence of  Task-Driven Autonomous Agents (BabyAGI, AutoGPT) it underlines what customers desire. We want one model that will become our go-to destination, a direct competition to Googling stuff. I want to be familiar with a model and its capabilities, especially as they soon will become more autonomous (access to your text messages, calendar, email... maybe banking account) to complete given tasks, a true digital personal assistant/concierge. Things that Google Assistant a few years ago tried to deliver but failed in most cases.

What we don't want (B2C)

Chatbots.... that are limited to each company. When I look for new shoes I want my AI model to compare all available models on the market, maybe even second-hand, but I don't want to spend hours going to each company and explaining to each chatbot on their website what I'm looking for. This approach is doomed as ineffective, and also a direct opposition to spontaneous shopping (on which eCommerce is built - "buy it is on a promotion, if you don't like send it back").

There are some exceptions where voice/text AI Assistant might work:

  • Market monopoly - so you are "the place" for looking for a particular need, and your customers are never considering comparing your offer to the competition. Ex. or are probably in a position to launch such products with some success.
  • B2B - if you have to use ex. BigQuery or Google Analytics as your data source it is ok to limit yourself to "AI Assistant" that will utilize only those products to help you ex. create SQL queries. This goes for business services of ex. Google/Microsoft/Atlassian.

What to do (as a company)

If you are a company looking into AI...

Enhance already existing features by "AI dust" - ex. or  Notion and dozens of others are doing this the right way. Find the most time-consuming part and try to make it better with the help of automation and AI (ex. Design System in Figma). This is a quick-win strategy.

Step aside and look at your company and its processes/services/value proposition etc. and consider how AI fits into this segment. Being to disrupt your own yard of existing products/services is very hard and only a few companies succeeded in it. Meta is doing this now with the push for VR/AR and only time will tell.

Prepare that soon "bots" can mean your client's Task-Driven Autonomous Agent that will order and behave on their task (ex. ordering shoes on a person ask), this can lead to interesting challenges, I think it will soon be a domain similar to SEO to make sure ex. BabyAGI or AutoGPT can use your shop/service.

Internal tooling - Github Copilot or many other tools that can help your employees boost their effectiveness.

But please, please, don't do another "AI Assistant" limited to your website and call it "AI revolution", it is just a bit better chatbot and we already had hype for those.

I don't think it is a problem only with Zalando and its "AI fashion assistant" (especially as it isn't yet available), but this was the most recent announcement that directly showcase slapping AI hype on a chatbot, when there is so much more happening with AI, even in fashion (ex. AI models) or even AR fitting that Zalando is doing.