You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Commit 5531278
Browse filesBrowse the repository at this point in the historyBrowse files
What happens to InnerSource when producing a contribution is no longer the hard part?
476
+
477
+
AI can create code, tests, documentation and pull requests in minutes. That shifts the constraint. Review time, ownership, trust and shared context become even more precious. Adding agents to today’s operating model will not automatically create better collaboration. It may simply create a much larger queue.
478
+
479
+
In this talk I will ask what comes next for InnerSource and offer a socio-technical answer. Drawing on more than 30 years helping large and regulated organisations improve how their engineering systems work, I will follow an AI-assisted contribution from idea to production. Along the way, I will look at the relationships between contributors, trusted committers, platforms and product owners, and show where the familiar model may need to evolve.
480
+
481
+
I will introduce five practical ideas for agent-ready InnerSource:
482
+
1. Publish intent and contribution boundaries before code.
483
+
2. Make ownership and contribution paths easy for people and agents to discover.
484
+
3. Put repeatable governance into paved-road workflows.
485
+
4. Protect maintainer attention through progressive trust and clear evidence.
486
+
5. Measure reuse, learning and flow rather than pull-request volume.
487
+
488
+
This is not a talk about AI replacing communities. It is about why community becomes more important when implementation is cheap. The people who decide what should exist, who can change it and how the organisation learns together are still at the heart of the system.
489
+
490
+
Attendees will leave with a simple diagnostic they can use on a repository, a platform or an enterprise InnerSource programme the next day.
Many organizations recognize the value of InnerSource in breaking down silos, promoting code reuse, and fostering collaboration. Yet, adopting InnerSource at scale remains challenging. Engineers struggle to discover reusable projects, understand unfamiliar codebases, navigate contribution processes, and build confidence when contributing outside their teams. At the same time, InnerSource Champions, Trusted Committers, and Maintainers face the ongoing challenge of onboarding contributors, maintaining documentation, mentoring communities, and measuring the health and impact of their InnerSource programs. These organizational and technical barriers often slow adoption long before engineering challenges arise.
505
+
506
+
This talk explores how Artificial Intelligence can accelerate the adoption of InnerSource by reducing friction at every stage of the organizational adoption journey. Beyond AI-assisted coding, it examines how AI can increase awareness of InnerSource, help engineers discover reusable projects, simplify onboarding, identify contribution opportunities, and provide repository-specific guidance for first-time contributors. It also explores how AI empowers InnerSource Champions, Trusted Committers, and Maintainers by automating repetitive tasks, improving documentation, supporting mentoring, surfacing community insights, and measuring program health. By lowering the barriers to participation while strengthening the capabilities of those who lead and sustain InnerSource communities, AI becomes more than a developer productivity tool—it becomes an enabler of organization-wide collaboration and a catalyst for sustainable InnerSource adoption.
507
+
508
+
The session also examines the organizational capabilities needed to make AI effective, including trusted repository-specific context, transparent governance, and well-defined InnerSource patterns. Through practical examples and an incremental adoption roadmap, attendees will learn how AI can increase contributor engagement, improve collaboration across teams, and help organizations scale their InnerSource programs sustainably.
509
+
510
+
The future of InnerSource is not simply AI-assisted development—it is AI-powered collaboration that enables every engineer to become a confident contributor and every repository to become a shared organizational asset.
511
+
<tdclass="UTC">UTC 13:00</td>
512
+
</div>
513
+
</td>
514
+
515
+
516
+
</tr>
517
+
<tr>
518
+
<td class="time">14:30 CET</td>
470
519
<tdclass="author w-70"> <b><ahref="#Gilles_Gravier">Gilles Gravier </a></b><br>(Geneva State Administration IT Services)<br>
471
520
Implementing InnerSource in a Government - Where we are after 2 years
Working in a government is tough. Lots of red tape. People not ready for change. In this session we'll see how things have progressed after 2 years of trying (yes, trying) to set up an ISPO in the Geneva Canton IT Organization. It's going to be fun!
476
-
<tdclass="UTC">UTC 12:30</td>
525
+
<tdclass="UTC">UTC 13:30</td>
477
526
</div>
478
527
</td>
479
528
480
529
481
530
</tr>
482
531
<tr>
483
-
<td class="time">14:00 CET</td>
532
+
<td class="time">15:00 CET</td>
484
533
<tdclass="author w-70"> <b><ahref="#JosepOriol_Ayats">JosepOriol Ayats </a></a></b><br>(AXA Group Operations)<br>
@@ -492,14 +541,14 @@ What if we approached AI usage the InnerSource way?
492
541
In this talk, we'll explore how open collaboration practices can dramatically reduce token consumption while improving engineering outcomes. We'll share practical techniques to reuse prompts, standardize context, build shared knowledge assets, and create community-maintained AI workflows that help teams avoid duplicated effort.
493
542
By treating prompts, instructions, and AI practices as reusable InnerSource assets, organizations can save up to 90% of their token budget while enabling developers to learn from each other, scale best practices, and become more effective together.
494
543
Join us to discover how InnerSource principles can turn token optimization from an individual hack into a community-driven advantage.
@@ -508,20 +557,20 @@ Join us to discover how InnerSource principles can turn token optimization from
508
557
Surely we want to increase the number of PRs, to show that InnerSource is successful, or not? How about minimising the time it takes to review PRs? What about increasing the number of projects that are InnerSource-ready?
509
558
510
559
In this talk we will look at ways to measure InnerSource initiatives. We'll investigate trivial to implement but often misleading metrics. We'll show the importance of understanding the purpose of establishing InnerSource best practices to drive metric definitions. We'll also look at what surveys can uncover - and where communicating with people in 1-on-1 settings can help reveal issues before they grow into mountains blocking collaboration.
@@ -575,7 +624,7 @@ Here's the thing: some of the most durable outputs of InnerSource aren't just sh
575
624
And here's the twist: it's also exactly what AI runs on. That messy human context is what makes an AI assistant useful rather than generic. AI consumes organizational knowledge. It doesn't produce it. Worse, it can quietly degrade the tissue itself: when the questions that used to go to a colleague go to a chatbot instead, the collaboration and the relationships that generated the answers stop forming. Stack Overflow's collapse in new questions is the public preview. The wave of AI-generated pull requests hitting open source maintainers is another.
576
625
577
626
InnerSource practices do the opposite. They are one of the few deliberate ways organizations build shared knowledge and cross-team relationships. So this talk makes the case for InnerSource communities as stewards of the connective tissue AI depends on and erodes: what to hand over, what to hold onto as distinctly human, and how to answer the inevitable ""can't AI just do that?""
578
-
<tdclass="UTC">UTC 15:30</td>
627
+
<tdclass="UTC">UTC 16:30</td>
579
628
</div>
580
629
</td>
581
630
@@ -593,7 +642,7 @@ InnerSource has proven to be a success when working with highly reusable framewo
593
642
But, when you work in a holding-group, you see that companies are struggling to keep up and repeating the same mistakes over and over again. Which made us wonder, what if infrastructure itself could also become a reusable framework? But not only as single components, but a whole suite that is extensible and customizable to every company's needs.
594
643
595
644
In this talk, we'll explore how GitOps can turn shared infrastructure repos into genuine InnerSource projects. Using pull requests as the unit of contribution, Trusted Committers encoded in CODEOWNERS, policy-as-code guardrails instead of approval boards, and a control plane where ""merged"" means ""deployed.""
596
-
<tdclass="UTC">UTC 16:00</td>
645
+
<tdclass="UTC">UTC 17:00</td>
597
646
</div>
598
647
</td>
599
648
@@ -607,7 +656,7 @@ In this talk, we'll explore how GitOps can turn shared infrastructure repos into
This session will cover a set of practices that I've seen, and that should not be repeated(!), after participating in and championing InnerSource within my company. By sharing our mistakes, attendees can hopefully achieve success more quickly in their own InnerSource programs. Let our pain be to your benefit!
610
-
<tdclass="UTC">UTC 16:30</td>
659
+
<tdclass="UTC">UTC 17:30</td>
611
660
</div>
612
661
</td>
613
662
@@ -631,7 +680,7 @@ In this talk, we will go beyond the hype and dive deep into:
631
680
• Guidelines for contributors
632
681
633
682
You will leave this session with a clear, data-backed framework for understanding and improving your team's AI adoption strategy. Instead of focusing on just the tools, you'll learn how to identify and address the systemic issues that are holding you back, and how to create an environment where AI can truly amplify your team's success.
634
-
<tdclass="UTC">UTC 17:00</td>
683
+
<tdclass="UTC">UTC 18:00</td>
635
684
</div>
636
685
</td>
637
686
@@ -649,7 +698,7 @@ Open source and InnerSource are often treated as two distinct operating models:
649
698
This panel brings together industry practitioners working across open source, InnerSource, OSPOs, engineering governance, and enterprise technology to explore what each model can learn from the other. We will discuss how open source practices such as transparent decision-making, contributor pathways, maintainership, and community building can strengthen internal collaboration, and how enterprise InnerSource experience can help organizations become better open source participants.
650
699
651
700
The discussion will also tackle the difficult questions: When should an organization contribute upstream, maintain an internal extension, or fork? What processes, controls, incentives, and organizational structures unintentionally prevent effective InnerSource? And how can companies build a continuum where code and ideas can move more naturally between internal and external communities?
652
-
<tdclass="UTC">UTC 17:30</td>
701
+
<tdclass="UTC">UTC 18:30</td>
653
702
</div>
654
703
</td>
655
704
@@ -671,7 +720,7 @@ Everyone can "vibe-code/create" their own solutions now, does this mean reuse is
671
720
The bar for contributing to other projects has been lowered, but what does this mean for maintainers?
672
721
673
722
In this talk a seasoned InnerSource Officer will bring food for thought on how AI can impact the collaboration style prescribed by InnerSource, for the best of for the worst, and share ideas on how to navigate this ever-changing terrain. We will also look into intrinsic benefits we expect to have with AI, but also, where a human push is still needed.
674
-
<tdclass="UTC">UTC 18:30</td>
723
+
<tdclass="UTC">UTC 19:30</td>
675
724
</div>
676
725
</td>
677
726
@@ -688,7 +737,7 @@ In this talk a seasoned InnerSource Officer will bring food for thought on how A
688
737
What if scaling AI isn't really about adoption, but about what a community creates once people start using it together? Within a large U.S.-based financial services firm, 94% GitHub Copilot adoption and 78% agent-written code were only part of the story. The bigger outcome was turning individual experimentation into repeatable, discoverable patterns through shared skills, prompt, and harness libraries.
689
738
690
739
By applying InnerSource principles alongside grassroots engagement, storytelling, and community building, teams could contribute what worked, learn from one another, and build on proven approaches instead of continually starting from scratch. This session explores how AI adoption became a catalyst for a shared collective of knowledge—and how measuring reuse, contribution, and impact can tell a much more meaningful story than adoption alone.
691
-
<tdclass="UTC">UTC 19:00</td>
740
+
<tdclass="UTC">UTC 20:00</td>
692
741
</div>
693
742
</td>
694
743
@@ -709,7 +758,7 @@ We'll cover what Skills and MCP servers are. How to create them (with some simpl
709
758
Attendees should walk away with a good handle on how to make a skill and MCP server, and when. More importantly, how we can reliably share with easy to maintain repositories in GIT.
710
759
711
760
The sharing aspect of skills truly leads to democratization of Gen AI improvements and drives at the core of Inner Sourcing.
712
-
<tdclass="UTC">UTC 19:30</td>
761
+
<tdclass="UTC">UTC 20:30</td>
713
762
</div>
714
763
</td>
715
764
@@ -734,7 +783,7 @@ Preliminary session outline:
734
783
3. How do you get started? Choose one measurable use case, connect a small set of authoritative information sources, define access and ownership rules, launch a simple question-and-answer experience, and create a feedback loop that turns unanswered questions into durable organizational knowledge.
735
784
4. How do you measure success? Track faster answers, increased project reuse, greater contributor participation, fewer repeated questions, and less time spent locating information or experts.
736
785
737
-
<tdclass="UTC">UTC 20:00</td>
786
+
<tdclass="UTC">UTC 21:00</td>
738
787
</div>
739
788
</td>
740
789
@@ -755,7 +804,7 @@ This talk explores the next frontier of InnerSource: the shift from sharing code
755
804
What happens when expertise can be discovered, reused, improved, and composed like software?
756
805
757
806
The next wave of InnerSource isn't about code reuse. It's about agent reuse.
758
-
<tdclass="UTC">UTC 20:30</td>
807
+
<tdclass="UTC">UTC 21:30</td>
759
808
</div>
760
809
</td>
761
810
@@ -771,7 +820,7 @@ The next wave of InnerSource isn't about code reuse. It's about agent reuse.
771
820
InnerSource is often associated with opening source code internally so that teams can discover, reuse and contribute to each other's work. But in enterprise software, source code is only one part of what teams need to build and operate a product. There is also a lot of knowledge around privacy, security, compliance, architecture and operational practices that can be difficult for other teams to discover and reuse.
772
821
773
822
I have seen this challenge firsthand in privacy engineering. A product may be considered compliant with a privacy requirement, but when a customer, auditor or internal stakeholder asks, "How?", finding the answer can be much harder. The information may be spread across product documentation, architecture decisions, privacy assessments and the people who have worked on the product. This talk explores whether we can apply InnerSource principles such as transparency, discoverability, reuse and contribution to this problem. What if contributors were expected not only to contribute code, but also to make the privacy implications of their work and the "how" behind compliance understandable and discoverable? I will share some ideas for what an InnerSource approach to privacy compliance could look like and invite the community to think about where else InnerSource can go beyond code.
774
-
<tdclass="UTC">UTC 21:00</td>
823
+
<tdclass="UTC">UTC 22:00</td>
775
824
</div>
776
825
</td>
777
826
</tr>
@@ -788,7 +837,7 @@ For most of computing history, building software meant negotiating with scarcity
788
837
But when you can build almost anything, the hardest question is no longer *how*. It is *what...and why*.
789
838
790
839
In this keynote, Scott Hanselman explores what happens when creation becomes abundant. Through stories, code, and practical examples, he examines the new skills that matter: judgment, taste, curiosity, empathy, and the discipline to understand the problem before generating the solution. The future will not belong simply to those who can build the most. It will belong to those who choose carefully, build responsibly, and create things genuinely worth having together.
0 commit comments