Founded by one of the early pioneers of observability, Honeycomb.io, which previously shied away from adding AI capabilities to its products, has embarked on a natural language query driven by generative AI.
Charity Majors, CTO and founder of Honeycomb, is widely credited with coining the term. observability Demonstrates a holistic understanding of complex distributed systems with custom queries.She commonly describes herself as ‘salty’ AIOps Machine learning capabilities of IT operations tools. But this week, Honeycomb revealed that it will be incorporating his AI into its products for the first time via OpenAI. GPT API.
This new partnership is the basis for an experimental Honeycomb feature called Query Assistant. natural language query It is based on observability data and is now available free of charge to all customers.This feature can be turned off by vigilant teams data privacyAccording to the company’s press release, no user data is passively sent to OpenAI.
The TechTarget editorial caught up with Majors this week to find out why the value of AI in the case of generative AI in GPT differs in her view from the value of AIOps.
Does Honeycomb have its own query language? How were people querying before this?
Charity Majors: This used to be a graphical query builder, but honestly it would take a long time to do it manually. There’s also something called BubbleUp, although it’s not AI or ML, but it’s more effective at finding problems. If you’re sending data with some kind of intention, humans can draw bubbles around areas of concern on the graph. In the background it compares the inside of the bubble with the outside of the bubble, sorts them and compares them. This will show on top all the points that differ from what you are interested in. It’s this little loop that sums up his 99.9% of all system debugging: “This is what I care about and it makes sense.” What’s different about it? Then the machine immediately said, “These errors are all about devices from this version of Android, so they are different. This build id uses the language pack for this app in this region.” Tell you. It’s usually easy to tell what the bug is. If you can’t find them, it may take days or weeks to figure out what’s going on.
What do natural language queries add to Honeycomm for customers?
Majors: It’s like a great equalizer. Query Builder works best if you have a good understanding of SQL and a good understanding of your data.but [Query Assistant] People who aren’t very sophisticated developers or even engineers are asking, “I just deployed something. What’s slow or what has changed?” or “Which users have the most bugs and errors?” ?” The tool is completely out of the way. This isn’t for expert level stuff, but it’s great for getting started.
I am on record as being incredibly cranky about most AI products on the market. But I am really excited about this.
Charity MajorCTO and Founder of Honeycom.io
Honeycomb was necessarily a complex tool. You can ask questions to understand the system and not be tied to custom metrics or anything like that. But our developer tools philosophy is that we should strive to get out of your way. You should try to focus only on the question you are asking, the problem you are trying to solve. Don’t even try to find a tool trying to solve it as a middleman. This is such a beautiful move forward. Anyone can tweak the query. It also threw AI at the salad and just said, “Wow! I’ve been recorded as being incredibly cranky about most AI products on the market. But I’m really excited about this one.” It helps engineers improve their work in meaningful ways.
What’s the difference between Query Assistant and “Throw AI in your salad”?
Majors: My big complaint with AIOps is that it’s usually a solution that shouldn’t be needed to a problem that shouldn’t exist.one of the things [AIOps vendors] Often, “If I’m inundated with emails from alerts, I’ll let you know which ones are important.” But you don’t have to send millions of alerts. they are useless. What they’re basically saying is, “OK, humans can’t make sense of all this noise anymore.” So let the machine do the work. ’ But it is even worse. Because it’s twice as frustrating when your machine alerts you in the middle of the night and you don’t know why.
This is where Bayesian math comes into play. He has a 90% success rate, even if he happens 1 time in a million. The most costly thing reliability engineering can do is false positives.You’re just training people not to pay attention to alerts – it wears them out. And when something goes wrong, the whole point of this [tech] The industry understands your system. We are not outsourcing the understanding. The machine will find the signal, process the numbers and tell you if there are spikes or blips, but it will never tell you if it was good. I don’t know if that is intentional. Only humans can give meaning to things, and what most AIOps are doing is trying to take the meaning-making stuff from humans and give it to machines. Because eventually humans will come and you’ll have to figure it out, and you’re going to hold them back and cripple them.
What else can Honeycomm do with natural language queries in the future?
Majors: No way to keep state [in queries]This is one drawback. Queries cannot be iterated. It’s a fresh slate every time. For those just getting started with MySQL datasets, “Here are some very interesting things people are using. Systems like yours.
There are some very interesting [around shared knowledge] I can’t wait to go in. If I’m working in a corner of the system, I’m well aware of it, but if I’m debugging, I need to debug the whole system. And most of the knowledge about that system is not in my brain, but bits and pieces of it are in your brain, mine, Caroline’s brain, Ben’s brain. If the AI is a little angel or demon sitting on your shoulder and suggesting things, having access to it means, “Here’s what other people in your system are doing to interact with it.” I tell you By looking at experts, other experts. If that information is provided to you on a platter, you can be smarter and more effective at your job because you’re tapping into the wisdom of the crowd.
Beth Pariseau, Senior News Writer at TechTarget, is an award-winning veteran of IT journalism.she can be reached at [email protected] Or on Twitter @PariseauTT.
