At HP, modernization is not a luxury, but rather the key to many of their successes as an industry leader in technology. With a culture focused on innovation and invention, it’s not surprising that much of their work focused on developer experience includes experimentation with the latest technologies and frameworks to help them maximize productivity and transform the software development lifecycle.
As part of their AI initiatives, HP has been working with integrating the Model Context Protocol into their vision for an agentic SDLC, including using the Stack Overflow Internal MCP Server. Per Evan Scheessele, HP Distinguished Technologist, Stack Overflow’s MCP server has helped HP conceptualize bridges between knowledge silos among developers and their teams. It has also proven itself to be a fully functional, successful MCP server—one that serves as proof of concept for MCP as a framework, and an example against which to compare as they build their own MCP servers for proprietary business solutions.
Ultimately, HP’s experimentation with Stack Overflow’s MCP server is paving the way for transforming the developer role at HP into one that is “directive,” where employees are able to guide work driven by AI to increase their technical teams’ productivity, lower toil, and provide more time for collaboration.
Innovation at any company means opening up processes for transformation, allowing frameworks and mindsets to be shifted. At HP, much of their experimentation is led with an “eyes-open” strategy, aiming to stay keenly informed on the newest technologies and the strategies surrounding them. However, this experimentation must balance the need for invention and modernization with the strict governance and accountability that makes HP one of the most trusted brands in the tech industry. Any introduction of new technologies, including AI, into the company requires meticulous work to ensure security and quality. With this in mind, Evan Scheessele’s Developer Experience and Applied-AI Services team leads several experimentation efforts at HP, modernizing the software development lifecycle that helps keep HP at the cutting edge.
HP’s rigorous testing and high-optics governance of any AI solution meant that their team would need to do their due diligence when integrating the Stack Overflow MCP server into potential workflows. “AI governance is a big deal in the enterprise,” Evan said, discussing his team’s experimentation with MCP servers. “We’re trying to be rigorous. We trust but verify. We’re building a whole verification process.”
Even with this critical eye towards security and governance, the seeds for HP’s use of Stack Overflow’s MCP had been planted far before the Model Context Protocol was first announced. HP, in partnership with Stack Overflow, had been discussing the possibility of a knowledge ingestion connector between Stack Overflow’s enterprise product and GitHub’s Copilot. This connector would allow important context from HP’s internal knowledge base to be accessible for their AI coding assistants. But then, “We started reading about MCP and the writing was on the wall,” Evan Scheessele said.
MCP became a buzzy topic at HP, showing that there was excitement amongst technical teams at the possibilities MCP could bring to their AI systems. It became clear to Evan and his team that MCP servers had the potential to rise above other frameworks for agentic AI. “It’s a means to an end which is leaning into the next phase of AI software development—agentic SDLC, real workflows of agent-powered automation,” said Evan.
Like when the Developer Experience and Applied-AI team first began to think about integrating Stack Overflow’s enterprise knowledge solution into their AI workflows, they saw Stack Overflow’s MCP server as an opportunity to bring necessary organizational context to AI agents, breaking down the barriers to efficiency and collaboration caused by tribal knowledge. “In the enterprise you have a lot of enterprise global context…our security and privacy folks have these corporate level global contexts that are unique to how we do our code and our solutions,” Evan explained. “MCP is a legitimate means to bring that context to agent platforms.”
With a company that includes over 4,000 developers, HP’s ability to share knowledge is pivotal to the cross-functional abilities of their teams. Within their technical teams, many developers rely on institutional knowledge. However, this knowledge is often shared solely between close collaborators and team members, making information access slow-moving and inefficient. This implicit but crucial knowledge, sectioned off amongst teams, causes increased load on developers who have to store and share this context all on their own, creating barriers to cross-team collaboration and slowing productivity. “No one developer or program manager or product manager has all of that context,” Evan said. “And more and more I think it’s obvious they shouldn’t have to try to have it all in their head.”
While Stack Overflow’s enterprise product, Stack Internal, has helped HP by creating a central knowledge library for their technical teams, the advent of AI coding agents has compounded the challenges of surfacing knowledge in an organization of HP’s size. Coding agents whose models are trained on publicly available data are unable to access the proprietary knowledge and context necessary to improve productivity for developers at a company like HP. Because of this, AI coding agents are limited to generating simple code. “My goal is to really turn that on its head,” Evan shared. The Developer Experience and Applied-AI team at HP needed to bring this necessary enterprise context to their agents in a way that was manageable, living, and scalable for an organization with thousands of employees. They needed to “make those sources of truth systematically addressable [through] a process like an agentic engine or an agentic SDLC.”
“It’s a very special case with Stack Overflow, because it’s a democratized source of well-validated, upvoted and accepted answers. [It’s] a robust source of truth. That’s a key piece of the puzzle.”
For Evan and his team, the Model Context Protocol is showing itself as the “best path forward” for the next stage of the AI-driven software development lifecycle. “I’m thinking about MCP [as a way to] enable our coding agents and our SDLC workflow agents,” Evan said. “They have all of that enterprise context they need to do their job […] to amplify and parallelize our SDLC.”
As the first corporate knowledge MCP that HP has worked with, Stack Overflow’s MCP has proven to be ideal for question-answering and integrating information directly into the IDEs of developers. “It’s a very special case with Stack Overflow,” Evan explained, “because it’s a democratized source of well-validated, upvoted and accepted answers. [It’s] a robust source of truth. That’s a key piece of the puzzle.”
A key strength of Stack Overflow’s MCP server is its natural language and structure, which lends itself well to the ways humans search and discover knowledge. Although still in the discovery phase of working with MCP servers, the Developer Experience and Applied-AI team at HP has found Stack Overflow’s MCP to be “a very exciting precedent.”
“It’s optimized around asking a question and getting an answer,” Evan shared, “And the first time I touched it, it worked the way it felt like it should.”
Part of HP’s work with MCP servers was also to learn from them as they built their own MCP “broker”—an internal MCP that would serve as a catalogue of all other MCP servers available to developers. By creating this, they hoped to not overwhelm their AI agents, while also battling tool fatigue for developers by creating one single connector.
For HP, Stack Overflow’s MCP has been a “proof point” for the success of MCP as a framework. “My team building [HP’s MCP] has learned an immense amount by understanding what the experience is like using [Stack Overflow’s MCP],” Evan Scheessele shared, “So in one way for us it’s a precedent… ‘Oh, this is what it looks like when it feels right’ […] Yours just seems to work perfectly.”
When building and testing their own internal MCP server, HP has used Stack Overflow’s MCP as part of their trial runs. Its seamless integration into agentic tools and ease of use for developers has made the Stack Overflow MCP server a validator of MCP as an overall framework, as well as a validator of HP’s internal MCP broker experiment. “It’s validating our roadmap of bringing multiple MCPs together to bring as much useful context together,” Scheessele said of Stack Overflow’s MCP. “I’m excited about it.”
HP’s experimentations with MCP—and agentic AI as a whole—pave the way for a modernized software development lifecycle. “It’s about developer experience,” Evan explained. “It’s a different experience when you’re using an AI assistant at a high level.”
For Evan, a fully functional AI agent—one that has access to the necessary technical context and knowledge—will allow developers to think more strategically and work more productively. “There’s a whole transformation of what it means to be a developer. It’s exciting because you’re directing the work of, potentially, a myriad of simultaneous parallel agents doing a lot of work,” Evan said. “So [developers’] actual productivity is off the charts.”
As Evan’s team continues to test frameworks like MCP for their agentic workflows, balancing innovation and invention with security and due diligence, they’re optimistic about what technologies like Stack Overflow’s MCP Server can bring to the software development lifecycle. “We’re seeing that we have enthusiasts who are finding that bleeding edge and unlocking immense amounts of productivity,” Evan said. “So we want to do that for everybody.”
For a company like HP, technological evolution is what makes them a long-standing leader of the industry. That’s why organizational transformation and modernization are pivotal for their strategic business goals. “I’m a believer that [MCP] is one of those enablers,” said Evan. “[We believe we’ll] be a hyper-optimized software company through this work.”
