The invincible march of AI continues to collect speed. Analyst Gartner recently estimated that half of all commercial decisions would be fully automated or at least partially promoted by AI agents within the next two years.
Also: 4 ways to convert AI into their business gains
Some organizations have used more than others. Four business leaders have discovered AI, who recently learned at a media roundateable event in Snomplake Summit 2025 in San Francisco. What did they say here.
1. What is my cloud strategy?
Astrazheneka’s main venture architect Ven Filin-Mathues explained how his organization is leading AI implementation in many areas.
Pharma giant has developed an AI-competent research assistant that enhances the productivity of scientific researchers by focusing on scientific methods breeding and development of new drugs.
Astrazane’s partner with key educational institutions such as Stanford University to run agent AI experiments.
“We are wondering how you may have a team of agents who can support traditional scientists doing their research,” said Philin-Maths.
Also: 4 ways can be favorable and flourished at the age of your organization AI
The company is also searching how to implement AI in commercial sectors. Astrazneneca 126 operates in markets, and serving those various places with materials is a complex challenge. This is where AI comes in.
“We have taken advantage of technology from AI’s point of view to automate information about the construction of marketing materials and the development of medicine,” he said, “he said.
While these experiments have highlighted the benefits of AI, they have also shown the importance of solid data foundation.
Also: Integrating AI begins with a strong data foundation. Here are the officers of 3 strategies employment
Filin-Mathues said that companies can solve problems with AI only if they have created a strong underlying cloud infrastructure.
“There are many cases of uses where the benefits are getting clear because we have gone on this journey,” he said.
“We are definitely in the era of making A-SAC decision. But the key to me is that you cannot forget those other underlying elements. You can’t be AI-first without cloud-first.”
2. Have I addressed the concerns of data governance?
Amit Patel, Chief Data Officer of Wholesale Banking in Trust, said that he learned two major lessons from rolling out AI use cases.
Number one was the importance of the underlying data foundation.
“As a bank, we have to prove, ‘Where has the data come from? Is this correct? Is it governed? Do I have a descent? Do I have Matadate? Do I have data quality checks?’ I have to prove those points to an external regulator, “he said.
“I can’t just release a large language model (LLM) in the wild, okay? And I can’t only indicate it at any source that I have. This is a ruled source. This is an authorized provision point.”
Also: It converts free Google Tool AI into your research assistant
Patel said that this focus on regulated sources helped clarify a common problem point for CDO: obtaining your data in order.
“Through that process, I have come to know that I do not have as much reliable sources as I would like to indicate,” he said. “I got to enable that foundation first, and then I can build on top.”
Patel said that the other thing he learned is that people who use AI at home believe that it will be easy to deploy LLM in an enterprise environment.
“It’s not so easy,” he said. “You must define the metadata to direct the model interpretations. And that process takes time.
Also: Is your business A-Taiyar? 5 ways to avoid falling back
Patel said that his team has addressed employees’ misunderstandings about the time of taking advantage of AI through hope-setting exercises.
“As we have started enabling cases of use, people have begun to understand that it is not as easy as a point-and-click process,” he said.
“While implementing technology is faster, it is still challenging, and it requires time and consideration that you put the rule and structure around AI before you enable it to work.”
3. What is the quality of my output?
Anhita Tafaviji, the main data and analytics officer at Snowflake, said that her team helps the tech company to develop A-Saksham products that use their customers.
However, Tafaviji said that his company does not just sell these products – the organization also gets to experiment with these techniques.
“The interesting thing about having a CDO in a data company is that I get the privilege of being the first customer of many of our products,” he said.
Tafvizi attracted attention to snowflake intelligence, a technique launched at the summit that allows business users to make data agents.
Too: Top 20 AI tools of 2025 – and to remember the number 1 thing when you use them
His team partnered with the product team to develop AI-competent assistants for the internal sales organization.
He admitted that implementing new AI devices leads to challenges, especially when it comes to balance the velocity of innovation with the requirements of governance.
An important idea is quality. As his team pushed the equipment to the sale team, he indicated important questions, such as “Is 95% quality quite good?”
Tafaviji advised other business leaders to think carefully about these challenges, as employees should rely on the output of AI experiment.
“Focus on quality has been important to us,” he said. “Right governance structures, access control, lineage, metadata and cementic models are also important. We consistently think of things that are in the form of stress between innovation and velocity.”
4. Have I considered unexpected benefits?
The main data and analytics officer of Finance Technology Specialist TS Imagine, Thomas Bodenski, said that his company has been using AI to reduce employees workload since October 2023.
However, while AI’s attention is often on automatic manual processes, their experiences suggest that business leaders should identify that technology also produces other benefits.
“Using AI is not just about reducing the effort,” he said. “You improve things faster, and also improve an incredible coverage.”
He explained how TS imagination buys data from expert vendors who send emails about upcoming product changes.
Also: 10 strategies Openai uses to create powerful AI agents – which you should also use
The company receives 100,000 of these emails in a year. Every email has to be read and its implications are understood. Traditionally, that work-intensive process has consumed two and a half full-time counterparts per year.
“It’s stressful because you can’t make mistakes,” he said. “If we remember the information in an email, our systems will go down. Thousands of traders cannot trade and thousands of risk managers cannot assess their risk, so it is potentially frightening.”
To avoid this scenario, Bodensky said that the company uses the AI ​​model of Snowflake to complete this time-intensive task.
“Now, we never miss the result,” he said. “They can do knowledge work rather than two and a half full -time equivalent manual data cures or entry.”
Bodensky said AI could also manage what was a weak location earlier: ensuring that the customer’s requests are settled on Saturday.
“No one has worked in those days. Now, AI is, and that customer will respond to inquiry and provide ticket to the right person,” he said.
Want more stories about AI? Sign up for innovationOur weekly newspapers.
