Every year around this time, tens of thousands of IT professionals prepare for a strenuous week in Las Vegas. And while the city sometimes known as “Adult Disneyland” is not short of attractions, for one week they are (mostly) the backdrop to what we are all in town for: basically, nerding out to our heart’s content on the state of The Cloud, and these days, IT in general.
Yes, it’s the AWS re:Invent conference!
If you go to the re:Invent website and click on Topics, you’ll see that the first item on the list is “AI/ML”. Then you’ll notice that the next item is “Analytics,” and then you realise that actually, the list is sorted alphabetically. But why let such boring details stand in the way of a good story? Regardless of how the AWS re:Invent team sorts the event agenda, it’s a pretty safe bet (even in Vegas!) that those topics will indeed be at the top of most attendees’ minds.
It’s worth reminding ourselves that the current wave of interest in AI (specifically, generative AI or GenAI) is barely two years old. At re:Invent last year, many of us thought that interest could not possibly grow any further — but sure enough, it has. What has changed is that we have moved on from the first phase of white-hot innovation and experimentation to seeing the first fruits of this technology in the real world — not technical innovation in a vacuum, but solving real, concrete business problems.
Early AI innovations at SnapLogic
We had a head start on other vendors, thanks to an in-house AI team led by SnapLogic’s Chief Scientist, Greg Benson. We already had AI features in our platform in those long-ago days before ChatGPT and launched our own SnapGPT (see what we did there?) integration copilot experience as early as March of 2023.
Launching early isn’t just about bragging rights; it also let us get our tech into the hands of customers, who are always the final judge of what we develop. That conversation is fast-moving and iterative, with feedback from real-world deployments going quickly back into the next development cycle. Early use cases were all around the “day one” experience, helping new users access documentation, build pipelines with assistance, document their own work, and understand the work of others through natural-language explanations.
We quickly moved to add more “day two” features, which would assist and accelerate the work of more experienced users: checking for best practices, refactoring for better performance, and generating SQL queries and expressions.
However, AI technology has advanced so quickly, that while chatbots and conversational interfaces make for a great demo, they are far from the only use case for GenAI and LLMs. And at the same time, it was becoming clear that not all of those early demos were achieving success, at scale and in production.
The biggest problem was getting reliable and continuous access to data. A demo could be built with sample data or a static snapshot, but to be useful in the real world, data has to flow from diverse sources in real time. In a messy IT architecture, assembled over time in response to different constraints, that flow is inevitably choppy and interrupted.
Building a robust data fabric for AI
That’s where we come in. SnapLogic has been connecting anything to everything, from the cloud to the ground, since day one. We were able to deliver in days what people — smart, competent people — had been struggling to achieve for months. The reason is not a new, magical AI model; we use AWS Bedrock to give us access to the best of what’s out there, so if anything, what we deliver there is a choice of models so that users can find the one that is the best fit for the job.
The way SnapLogic ensures the success of AI projects is precisely through the seamless data fabric that makes it easy to keep your AI-enabled feature fed with the best and freshest data, and plenty of it.
Once you are operating in that mode, it becomes obvious that AI is a lot more than a fancy front-end to your enterprise data — although that is of course very valuable in itself. But AI can also operate behind the scenes, in the shape of autonomous AI agents that seamlessly integrate into existing workflows.
This “headless” mode is very different from the more visible chatbots, but is even more powerful, with agents able to act semi-autonomously to deliver novel functionality, working with existing data and systems. Once again, SnapLogic is there to help, with our AgentCreator enabling everyone to build agents that scale affordably, integrate seamlessly with 1,000+ systems, and empower teams to quickly create value with AI.
Making AI and data accessible to everyone
When I say “everyone,” I do mean everyone. At our Integreat event in London just a couple of weeks ago, I led a room of 60 or 70 people through a hands-on workshop where they used SnapLogic’s GenAI capabilities to solve a series of real-world problems: analyse a bill, compare one bill to another, and write an email with your findings.
The group was about evenly split between existing SnapLogic users who were there specifically to try out the new AI capabilities, and people completely new to SnapLogic who were there out of curiosity or as part of their own evaluation of the platform. There was also a huge variety of titles and backgrounds, from individual data scientists to business teams, and even Brian Dummann, the Chief Digital Officer of AstraZeneca, who was one of our keynote speakers.
Everyone was able to finish the exercise (the LinkedIn badges are on their way, folks!) and more importantly, gain a much more concrete understanding of how all of the different components work together than you could get from any number of slides.
See it for your very eyes at re:Invent
If you’re at AWS re:Invent and you want to find out more about SnapLogic in general, and AgentCreator in particular, please do visit our booth during the week. We are also hosting a Lightning Talk on Monday at 5:30pm, hosted by Regina Goodman, SnapLogic’s Global Director of Solutions Architecture for GenAI, and would love to see you there. But if you’re not in town or you miss us, you can schedule a demo of AgentCreator any time.
