7 reasons you need Azure AI Search (even if you think you don't)
"I don't need Azure AI Search." Pluralsight's Amy Coughlin explains why this is often wrong, and the common fallacies around doing things the hard way.
Feb 13, 2024 • 4 Minute Read
- AI & Machine Learning
Ever heard the saying, "six hours of debugging can save you five minutes of reading documentation?" It's a saying almost as old as the profession itself. And yet, 50% of developers still spend time debugging anyway, and calls to "RTM" are littered all over StackOverflow (Yes, I'm using the polite version). The lesson here? Sometimes we do things the hard way, even when there's an easier, more efficient way.
The same thing can be said of not using Azure AI Search for your website and applications. Without it, you're spending more time and energy to give users an inferior user experience. In this article, I've listed out some arguments people use to not make their lives easier.
7 reasons for not using Azure AI Search
1. "Website users love a scavenger hunt."
No. They really don’t. Even if you have a low-traffic site, and you’re offering only a few dozen products, services, or experiences, your users expect to quickly find exactly what they are looking for – or they’re going to move on to another site that provides a professional search experience. Azure AI Search offers low-code provisioning and an SDK that makes it easy to integrate sophisticated search features with your website or app. You might want to give some though as to why your have a low-traffic site. Just sayin’.
2. "Three words: Bing, Google, DuckDuckGo."
That sort of feels like five words to me. But, I digress. You shouldn’t confuse Azure AI Search with search engines designed to for the world wide web (does anyone call it that anymore?) Azure AI Search is designed to help you curate and prepare data you control, and then wrap it in a professional search solution. However, if you enjoy leaving the discovery of your content and products to the whims of web search engines, then, by all means, you should skip consideration of Azure AI Search.
3. "ChatGPT is the same thing."
ChatGPT is an implementation of a large language model, or LLM. And, yes, it is a form of Artificial Intelligence, or AI. And, yes, you can apply ChatGPT to operate over data you curate and constraints you define. But ChatGPT is specifically designed to respond to your prompts, and respond as a human might by answering a question, showing you an image, or generating some code –- which may or may not be correct.
No amount of prompt engineering is going to populate an index that drives a cogent display products in a product catalog or documents in research library. You can, however, integrate Azure AI Search with the likes of ChatGPT, so why not both?
4. "I sell art. They’ll know it when they see it."
Sure. Sometimes, we can just browse through a brick-and-mortar gallery, and fall in love with an upside down can of tuna drifting down a magenta river to nowhere. But suppose you’re looking online for just the right wall art that includes wildflowers as far as the eye can see, or, perhaps, a castle in a brooding, dusky rain. Browsing through thousands of images just isn’t going to cut it – especially when you probably have very specific ideas around the media, size, artist, and price you are willing to pay.
An Azure AI Search solution can readily include Azure AI Vision to scan and tag the contents of each image, along with indexing all the other search parameters. Even at $10,000, “Tuna Float” might just match with its forever home.
5. "Coding from scratch is more fun."
As a software engineer, I hear ya. It is more fun. But I dare you. In fact, I double-dare you to use that truth as a rationale for not considering a PaaS offering such as Azure AI Search.
“Let’s see, I can code you up a good solution for that product catalog of tractor parts we’ve been tasked with building. It will take four more developers dedicated to the task, and we can demo something in a few months.”
Yeah, that will go over well.
Kudos go to people who propose and implement timely, solid solutions that work and are less expensive to deploy and maintain. Do I hear you saying, “But open source . . . “? Azure AI Search is an implementation of Lucene, the premier open-source search platform, combined with Microsoft’s proprietary algorithms that also integrate with other AI and machine learning services. Code that. If you dare.
6. "Promises of 'manually tagging content' attracts the best interns."
Ah, yes. But wouldn’t you rather they took coffee orders? They’d be free to do that if you used a couple of Azure AI Search’s in-line AI enhancements, called “skills.” Specifically, you can use Azure AI Vision to scan images to generate tags and captions – you can even apply OCR (optical character recognition) to both printed and hand-written text.
You can also apply Azure AI Language skills to extract key phrases or perform sentiment analysis. With all that automation, your interns might even be able to add smoothie delivery to their resumes.
7. "Our team isn’t ready for AI solutions."
I hate to tell you this, but if you’re thinking that way, you’re already a little late to the party. But let’s get you caught up with the “in” crowd. Check out my course, Introduction to Azure AI Search, along with my latest offering, Azure AI Search Learn by Doing, which includes just a little bit of lecture and a whole lot of labs!