Disclaimer

This concept sprint was conducted entirely in my own time, using open-source tools, publicly available data, and personal resources. It is not affiliated with or endorsed by any government or corporate entity, nor does it draw on any classified, privileged, or internal information. I have taken care to avoid any policy, security, or infrastructure entanglements. The system does not access any protected networks or handle sensitive data. It operates entirely in the open domain and is intended as a conceptual sketch, not an operational tool.

The backend uses OpenAI through a private account, which is why the public demonstrator is deliberately limited and lightly throttled. Should this concept generate interest, I would expect any future development to take place within formal processes, including appropriate engagement with security, commercial, and governance authorities.

Welcome

Welcome to the Defence Insight Garden, or "the DIG". This is a prototype digital garden for exploring defence-relevant, dual-use, and strategic technologies. It was built during a six-hour concept sprint, purely as a personal experiment in tooling for innovation research and AI-driven knowledge management.

A digital garden is a living, evolving knowledge space. Unlike a traditional website, it is not a fixed repository. It supports incomplete thoughts, ongoing revision, and cross-linking. Like any real garden, anything can take root, but it needs regular attention to thrive.

In this garden, a team of collaborative AI agents do the planting. Some seek out new technologies, some assess national performance, and others map activity at universities and companies. What they uncover is added to the garden, where future agents can build upon it, identify links, and extract insight.

Aside from this page, all others were generated by the AI collective. They were given a fixed pool of 36 technologies to work from. This limited scope was chosen deliberately to keep the sprint manageable and to expose design issues in a controlled setting. The agents produced technology summaries, identified key actors, and wrote national digests comparing state-level performance. For now, the digests remain restricted to these 36 topics.

The agents were developed and and operate within a structured knowledge environment. They share the same dataset (human readable in Obsidian), and they can respond in real time to new prompts or changes. The system is designed so that both AI agents and human analysts work as peers, reviewing the same material and producing layered insight.

The public version, available at dig.rizgarmella.com, uses a basic graph-style viewer. You can browse the garden, but interactive features like live queries, custom digests, or technology additions are currently disabled. Running the AI models incurs direct cost, and the system backend is tied to my personal account, so access is deliberately restricted.

You can start by exploring the United Kingdom or the United States.

There are bugs, omissions, and awkward artefacts throughout. For example, the USA and United States are treated as separate entities. These are known issues. Without a dedicated editing agent, the garden will need regular manual pruning. That is why the scope is currently small. Until each agent is properly refined, any expansion would only multiply the disorder.

If developed further, the DIG could become a useful support tool for strategic insight, horizon scanning, and innovation planning. Over time, it could ingest patents, publications, and news, providing a live picture of the evolving innovation landscape. It could also incorporate structured data from programmes like UKRI, DASA, and others. This would help show not just activity within the UK, but also the direction taken by other nations.

This is not a formal proposal, nor an operational tool. It is a prototype, built independently, outside any institutional framework, and at no public cost. Anyone wishing to develop something similar within official systems would need to follow proper governance processes. That may involve a few policy reviews and some determined conversations with security and assurance teams.

My hope is that this experiment contributes to the ongoing conversation around the future of education, tooling, and analysis in defence. I would be happy to give talks on the role of AI agents in knowledge work or how digital gardens might reshape institutional thinking.