Role of InnerSource the future of software development (e.g., with Agentic / Vibe Coding)

[dellagustin-sap]

Description

Generative Artificial Intelligence (GenAI, and currently, just AI), especially Large Language Models (LLM s), and Agentic AI have evolved to the point that, in several organizations, developers are expected to do Agentic Coding, also known as Vibe Coding. This indicated that the role of software developer is shifting partially from one that writes code, to one that provides instructions in natural language to a system that will implement the necessary changes. The responsibility of the developers is then shifting to one that oversees the work of (semi-)automated coding agents.

Some practices are also emerging where several automated agents, representing roles like Quality Engineering/Assurance (QE, QA), Project Manager, Frontend and Backend developers work in tandem to develop software and interact directly with supporting tools like Jira and GitHub.

The ease of generating software upon requirement also puts in question the role of software reuse in the future of the industry.

Taking all that into account, it is important to reflect on the status quo and possible future of software development and also the role that InnerSource takes in these future, as well as the role that AI has in implementing an InnerSource program / strategy in any organization.

(Potential) Impediments

• Context Size of available models vs (Code) Project Size and length of human-agent iterations

Potential Futures

• A future with less software component reuse - as developers may ask coding agents to rewrite any capabilities they want into their own component, instead of reusing existing components
  • Not expected to happen extensively with reuse of services that bring their own system of records, likely for compliance and internal company tech strategy and guidelines

Related Terms and Acronyms

• Agents
• Agent Teams
• Model Context Protocol, MCP
• Agent-to-Agent, A2A
• Skills

Ideas

Communication

Asynchronous

Synchronous (meetings and calls)

• 2026-03-02 ISPO WG Meeting

Opportunities

• Reusable InnerSource components could reduce the cost of AI
• AI fosters InnerSource: Better documented development processes are expected to improve performance of agentic coding
• InnerSource mitigating AI risks: Transparency and keeping things traceable and open for inspection could mitigate "AI slop"

Risks

• Short term success by added risk taking - People and teams taking more risk (e.g., releasing AI generated content with fewer human review)

Expected outcome

• New chapter on the book Managing InnerSource Projects about the topic
• Possibly one or more patterns for specific sub-parts of this challenge

Links and References

• Related standards and protocols
  • https://agents.md/
  • Model Context Protocol
  • Agent2Agent
• Related organizations
  • Agentic AI Foundation (AAIF)

[jeffabailey]

There’s an opportunity to write about integrating AI into your ISPO.

https://innersourcecommons.org/learn/books/managing-innersource-projects

[jeffabailey]

If you're in the TODO group, they're kicking off discussions about AI in OSPO context in #topic-oss-ai, there may be collab opportunities.

[jeffabailey]

Input from Fei Wan.

Fei Wan  [12:48 PM]
What I propose is to use agent skills to codify the enterprise standards/best practices etc. and use that in the enterprise process.

Here's the open standard for agent skills from Anthropic:

Overview - Agent Skills

My summit talk is on this topic. At that time the Agent Skills was not an open standard yet.

[12:51 PM]But the mechanism has been there. When I mention agent workflow in my talk, it's the agent skills.
[12:51 PM]Let me know if you have any questions. Happy to discuss this further.

https://youtu.be/kPX4-36Tma0