Here at Snowflake, we have a strong history of empowering our customers to optimize their processes with AI in the Snowflake platform and helping them achieve transformational results. Now we’re turning that focus inward to enable our teams across the entire organization to leverage generative AI to streamline their daily workflows, automate repetitive tasks and gain deeper insights from data. 

Let’s explore how two Snowflake employees from two distinct teams didn’t just adapt to AI but made it an integral part of their operations to improve their productivity, foster creativity and, ultimately, contribute to our company’s success.

Writing world-class interviews in a day, not weeks

Ryan Green, News Anchor and Senior Director of Program Content here at Snowflake, lives and breathes content creation. Whether he’s traveling the world to cover Snowflake events or hitting the show floor at Snowflake Summit interviewing industry leaders for Snowflake’s stand-alone news platform, Data Cloud Now, Ryan leads engaging discussions with Snowflake executives, customers and partners that viewers want to hear. But it’s not easy coming up with the questions and outlining the programs that keep people coming back for more — in fact, it’s a time-consuming process.

However, once Ryan started incorporating Google Gemini into his writing, his entire workflow changed in ways he wasn’t anticipating. One of the first major projects he used it for was to outline his interviews with thought leaders at this year’s Summit.

“I spoke to Gemini in natural language and was able to walk through my ethos of storytelling to build a prompt,” Ryan says. “And then, taking that one step further, I fed in my prior transcripts of public interviews as a resource guide. I also put in parameters on what Gemini was searching for to get better results. And within a day or two, I had the bones of my Summit conversations, which I could rework and prep for review.”

After the setup work with Gemini, Ryan was able to generate weeks of work in minutes. This gave him ample time to focus on refining his work to make his content as engaging as possible, with edits getting done within a day. “Now I can spend more of my time working on the real value-adds, like tone, delivery, pacing, strategy and turning conversations into bespoke content that’s hyperspecific and personal to each guest.”

Ryan conducted more than 40 interviews at this year’s Summit, and Gemini played a key role in building the structure of those discussions. But it’s important to note that Ryan didn’t have Gemini write his content for him. Instead, it augmented his workflows and got him from ideation to working drafts significantly faster. 

“It’s not replacing me, it’s just getting me to the starting line faster,” he says. “And it’s forever altered my approach to content creation, because I know I have control over the information it’s pulling into my drafts, and the writing is being done in a safe and secure way.” 

In light of the success that Ryan has had in upleveling his workflows with Gemini, he recommends that everyone across the marketing organization jump into figuring out how to implement generative AI into their own workflows. 

“I assure you, generative AI will unlock each and every one of your superpowers,” Ryan says. “Look at which of your tasks can benefit from the use of it, and then implement it to work faster and make better decisions.”

Building specialized tools to speed up content creation across multiple mediums

Across the organization, Chanin Nantasenamat, PhD, Senior Developer Advocate, has completely transformed his blog writing and content creation processes with the power of generative AI. Prior to using AI, it took Chanin around two weeks to produce a set of content, including blog posts, quickstarts, Jupyter/Snowflake Notebooks and YouTube videos, but he had a goal of reducing this timeframe by 10x to improve his team’s productivity, help them save time and scale their content creation abilities.

To do this, Chanin built a suite of tools with the open source web framework Streamlit, which completely overhauled his workflows:

• Blog generation: The Write-Blog tool allows him to feed in a Jupyter/Snowflake Notebook (and optionally, a YouTube URL) to convert it into a blog post in minutes, not days.

• Quickstart generation: Similar to the blog-post-generation app, the Write-Quickstarts tool creates quickstarts just as quickly by feeding the app a GitHub URL to a notebook and a YouTube URL (again, optional) for the corresponding tutorial, which adds additional context from the video.

• Quickstart/blog conversion: To expand the reach of existing content, the Write-Convert app converts between a quickstart and a blog post while preserving its content and code snippets.

• YouTube video description generation: Chanin also created the Write-Description app to generate the title, description and keywords for a YouTube video using only the video URL. This reduces the time needed to watch the video, manually write a summary and come up with an engaging, SEO-optimized title. The keywords it generates also help improve the video’s SEO.

While Chanin could have used existing LLMs for blog generation, the generated output didn’t meet Snowflake’s quality standards and required extensive rewriting and reworking to be as detailed and thorough as possible while also maintaining Snowflake’s tone and voice in the writing. Instead, Chanin chose to create his own app and trained it on posts he’d already written; he can even train it to match the writing style to follow the Snowflake brand style guide.

“This is helping us go from zero to a draft quickly, eliminating the tedious task of repeatedly starting from scratch to get that first draft ready,” Chanin says. No more staring at a blank slate; instead, leverage generative AI to quickly produce a draft.

It should be noted that the AI-generated content is based on a heavily documented Jupyter notebook as a “source of truth,” where code snippets and in-line explanations are the essence of the tutorial. The AI is essentially helping repurpose content to different formats that may have their own nuances.  

With the aid of his collection of apps, Chanin has significantly reduced his average content creation time to a few days or even less in some cases, including time for edits. In fact, in three to four months, the tools helped him author 14 content sets (each set consisting of a Snowflake Notebook, blog post, quickstart and YouTube tutorial video), totalling 56 pieces of content that would traditionally have taken several more months to complete.

To further empower others to create their own generative AI apps, Chanin has created a course in collaboration with DeepLearning.AI called “Fast Prototyping of GenAI Apps with Streamlit.”

Empowered to build faster and smarter

Seeing how our colleagues across the company are leveraging generative AI to work smarter and scale their processes has been inspiring. We’re learning from one another and continuing to embrace new, effective applications of AI to ensure we’re always growing, succeeding and at the forefront of change. And with less time spent on repetitive tasks and more time spent on building, developing and refining, we can bring more Snowflake innovation to our customers, faster.

To learn more about how to transform your processes and achieve success with generative AI and Snowflake, download the “Secrets of Gen AI Success” report or explore Snowflake’s AI solutions.