Unlocking Enterprise Transformation Through AI: Insights from IBM and Box
In the latest episode of the AI First Podcast,
This blog post synthesizes key insights and strategies highlighted in the conversation, delivering a roadmap for organizations aiming to make the best out of the episode’s valuable lessons.
(00:00) Introduction to IBM's AI-first transformation
Matt is helping IBM figure out how to deploy AI at scale “IBM is undergoing a massive transformation...figuring out “everyday AI” bringing AI into every part of IBM” is how Matt describes IBM’s approach. From a personal perspective, Matt has picked up coding again due to AI.
A central theme of the discussion was IBM’s vision to become “the most productive company” by transforming its workflows with a philosophy that predates the AI hype. Before generative AI dominated headlines, IBM had already adopted the mantra of “eliminate, simplify, automate” to drive business process transformation. Matt describes the vision for embedding AI into IBM’s business: “We like to say AI+ in everything”
Lyteson explained that the emphasis is first on reducing unnecessary processes, then simplifying core workflows, and finally applying automation where it drives value. Generative AI is now turbocharging this approach by unlocking entirely new capabilities instead of just accelerating existing tasks.
“Your first inclination is, yes, I'm going to find a way to do the same things I did before, only better faster... But I think quickly, then you start to realize... there are things that we didn't even really consider to do as part of our roles, and these are really the AI story that I think is really exciting.”
(07:30) Embedding AI into business operations and company culture
Jon mentioned the IBM approach is very comprehensive and all in. Matt mentioned, “IBM has an opportunity to put AI ontop of Hybrid cloud, and do it at scale from Day 1” Jon and Matt also emphasized the critical importance of governance when introducing AI technologies. The lessons learned from previous unregulated cloud adoption have steered IBM toward developing robust “enterprise guardrails.” These governance systems ensure AI tools are deployed ethically and intelligently across the company’s 280,000 employees.
Matt referred back to “Eliminate, simplify, automate” and how IBM is sourcing ideas from across the firm, in a lightweight manner, and implementing controls such as AI ethics review. A core principle is the human remains responsible for the output, and AI is a recommendation engine only.
Matt framed IBM’s approach as one that looks beyond individual use cases and instead examines workflows as comprehensive systems:
“We set up early guardrails to make sure we’re using a common enterprise platform built on our hybrid cloud. This lets us measure efficacy against workflows and apply governance in light, targeted ways while maintaining trust and ethical considerations.”
This focus on governance helps mitigate the risks of over-tooling, or what Herstein described as the “hammer and nail problem,” where every problem starts looking solvable by AI. The conversation reaffirmed the need for thoughtful application of AI, prioritizing areas that bring measurable, ethical, and scalable value.
(15:00) Generative AI's role in automating business processes
Another topic was IBM’s hybrid cloud strategy and its partnerships with leading SaaS providers like Box, SAP, ServiceNow, and Salesforce. Matt shared IBM’s intention to avoid building “by default” solutions, instead leveraging intentional designs that combine public cloud resources, data centers, and foundational tools like Red Hat OpenShift and IBM Cloud.
Highlighting Box as a key partner, Matt explained how Box’s robust integration with IBM’s AI technology (including Watson AI) is streamlining workflows:
“Most organizations aren’t in the business of building their own unstructured data sharing platform. Box embedding the Watson x platform underneath gives me a sense of trust and security. It means I don’t need to focus on where this data lives—Box becomes part of our core architecture.”
This collaborative approach allows IBMers to pull documents from Box, synthesize data with ERP systems, and efficiently create outputs—all within AI-enabled workflows that prioritize simplicity and security.
(21:00) Scaling AI adoption and overcoming challenges
“Today it’s a natural language prompt” said Matt mentioning “AskHr” as an example, where you need non deterministic answers, for example, for an employee who wants to understand remaining vacation days. How do you build these workflows in a way that is not overly burdensome.
“It forces us to re-examine the whole process” “It’s a fundamental rethinking”
Matt underscored the impact of thoughtful user experience design when rolling out AI-powered tools. Reflecting on lessons learned, he shared an anecdote about earlier missteps in designing user inputs for AI-driven workflows:
“We realized asking for 40 fields of data entry in a prompt window was a bad idea. Looking back, it seems obvious, but that’s what the team initially designed. You can’t just port web forms into AI interfaces—you need to rethink them.”
The takeaway was clear: AI interfaces shouldn’t burden users with redundant inputs. Instead, tools should leverage existing enterprise data to pre-fill fields and reduce unnecessary effort.
IBM has also taken intentional steps to acclimate employees to new AI-driven workflows, recognizing the anxiety that often accompanies technological change. One successful initiative involved developing “Ask IBM,” an internal tool trained on IBM’s intranet data to provide employees a safe environment for experimenting with generative AI:
“We did this intentionally for employees to engage with tools in a trusted way, specific to our organization. It helps them better understand AI's capabilities without feeling overwhelmed.”
(27:30) IBM-Box collaboration: Integrating hybrid cloud and AI
“We are now seeing 100’s of digital assistants” said Matt.
Jon asked about the evolution of the interface….”Most employees don’t go into our HR system anymore” said Matt.
Jon mentioned the evolution of proactive agents and workflow orchestration.
“I don’t think APIs are going away” “I still need to find the definitive source of truth for my HR data, for example”
Higher level and lower level Agents sit as the top layer, an abstraction layer to APIs.
Both Herstein and Lyteson asserted that organizations should move beyond simplistic measures like “hours saved” when evaluating the impact of AI tools. Instead, businesses should identify metrics that speak directly to financial leaders, such as unit cost reductions and workflow velocity improvements.
For instance, IBM has achieved significant reductions in hardware costs for employees by using AI-powered telemetry to optimize device allocation based on actual usage, rather than assumptions tied to job roles:
“We’re seeing such dramatic reductions in unit cost—almost below external benchmarks—while still giving high-capability machines to employees who need them most.”
Other examples included speeding up supplier brief creation from hours to minutes, translating real productivity gains into measurable value that CFOs and stakeholders can appreciate.
(36:00) Improving user experience with AI-powered solutions
Matt gave some examples of improved client experience and combining data across different sources. Matt spoke about metrics, such as hours saved. Value needs to come to bottom line, for example lower unit cost, to create an invoice, spin up a virtual machine. Reduction in unit costs for IBM machines/devices is an area where AI helps.
Another is flow velocity, how long does it take someone in procurement to prepare for a meeting with a supplier. The CFO can then review cost structure for different functions. Jon asks about culture and tips on change management. Matt mentioned the importance of coaching the team to trust the AI.
“We need to learn from each other. We created a library of prompts” Matt spoke about WatsonX challenge days, where people develop their own workflows. “How do I ask the AND question...” as a good way to learn what else AI can do.
Matt gave an example of summarizing a spreadsheet, and grouping categories..and asking AI ”what else can you do”
(50:30) Insights on AI, leadership, and the future of work at IBM
“Some people are gung ho, and others have more anxiety” “There are roles we will need to discover” said Matt
“People have to see the value” it needs to be “Good enough”
Jon asked Matt to summarize the impact of AI across three areas, Value, Culture and Employee experience.
- On value, Matt spoke about the importance of being clear and articulate, saving a few hours is not enough, it needs to be about increased workflow velocity or lower unit costs.
- On Culture “we need to learn together, what is the tip you can give me?”
- Finally, on experience “it’s got to be easier” on the importance of delivering a better user and human experience, particularly when it comes to working with agents.
Conclusion
The episode closes with Jon expressing gratitude for IBM’s decades-long partnership with Box and their shared commitment to driving technological innovation. He invites listeners to take actionable steps, challenging assumptions and staying curious as they navigate this bold AI era
Final Takeaways for Enterprise AI Adoption
This episode of the AI First Podcast offers guidance for businesses embracing AI:
-
Eliminate, simplify, automate: Lay the groundwork for transformation by removing inefficiencies before applying automation.
-
Governance and ethics: Build trusted and scalable AI systems with clear enterprise guardrails.
-
Integration: Leverage SaaS and hybrid cloud partnerships to unify fragmented data across ERPs, CRMs, and other systems.
-
User-centric design: Rethink AI interfaces to reduce data entry and leverage existing information.
-
Measure meaningful value: Move beyond “hours saved” to metrics like workflow velocity or unit cost impact.
Question for Box Community
How is your organization thinking about AI Adoption, do any of these takeaways resonate? Please share in the comments!