This post explores Laura’s real-world applications. For the original concept, see Building Laura. To try it yourself, follow the hands-on tutorial.

---

When I built Laura, I tested her on familiar ground - my undergraduate dissertation on the Spanish Civil War. Safe territory. I knew the material, could spot errors, could judge whether the output was genuinely useful or just plausible-sounding nonsense.

But the real test of any tool is whether it works outside the controlled environment where you built it. Over the past few months, Laura has been let loose on problems I never designed her for: tracking the provenance of a Victorian painting, supporting a family member’s transport planning research, preparing for job interviews, and - perhaps most mundanely - sorting out my own chaotic notes.

The surprise wasn’t that she worked. It was why she worked. The same basic structure - concepts, sources, questions, context - turns out to be universal infrastructure for thinking.

The Art Detective

A friend asked for help researching a painting. They’d acquired a Victorian-era work and wanted to understand its history - who painted it, who owned it, how it ended up where it did.

This is detective work. You start with fragments: a signature, a frame style, an auction house label on the back. You research family histories, exhibition catalogues, art dealer records. You build theories and test them against evidence. You hit dead ends and backtrack.

Laura didn’t know anything about Victorian art markets. She didn’t need to. What she knew was how to structure an investigation.

Within a few sessions, the vault contained:

• Source notes summarising auction records and catalogue entries
• Concept notes on the families who’d owned the painting across generations
• Question notes flagging gaps: “Verify whether the 1892 exhibition catalogue is accessible online”
• A research backlog prioritising next steps

If real life intervenes, the structure remains. My friend can pick the research back up in six months. The vault will be waiting - not a chat history to scroll through, but a structured knowledge base with clear next actions. They can resume immediately, as if no time had passed.

This is what persistent artefacts give you. Research that survives interruption.

Academic Research, Round Two

The Spanish Civil War vault was my original test case, but Laura has been tested with two other academic projects:

Transport planning research for a family member’s dissertation. Different domain, same structure. Concepts like induced demand and transit-oriented development. Source notes on peer-reviewed papers with proper citations. Maps of content connecting theoretical frameworks to case studies. The vault became infrastructure for thinking through complex arguments - not writing them, but scaffolding them.

Nordic influences on Scottish painting - a deep dive into artistic exchange between the Glasgow School and Scandinavian painters in the late nineteenth century. Laura identified key figures, traced exhibition connections, mapped shared approaches to light and landscape. Again: concepts, sources, questions, context.

My own dissertation revisited - the Spanish Civil War work I wrote thirty-five years ago. Different this time: I wasn’t an undergraduate with a deadline, but someone genuinely curious about historiographical debates I’d barely understood at twenty-one.

Three wildly different subjects. Same underlying structure. Same research methodology. Same tool.

Job Interview Prep

This one surprised me. I was preparing for a new role and realised that interview preparation is research.

Laura helped me build a vault covering:

• Industry context - market trends, competitive dynamics, recent news
• Case studies - how techniques mentioned in the job description had been applied elsewhere
• Concept notes - definitions and frameworks I might need to discuss
• Company research - structured notes on their strategy, challenges, and culture

The resulting vault wasn’t interview answers. It was the knowledge infrastructure that made confident answers possible. When you’ve systematically mapped the territory, conversation feels natural rather than rehearsed. Laura’s been able to scan my hand-written notes from chats with colleagues, flesh those out and link them in to the research concepts.

Housekeeping My Own Brain

The most unglamorous use case: sorting out my own notes.

I had scattered documents across multiple vaults - agile practices here, leadership reflections there, half-finished thinking everywhere. No consistent structure. No clear connections. The kind of knowledge entropy that accumulates when you’ve been working for decades without a system.

Laura helped me consolidate. Not by magically organising everything, but by applying the same structure that worked for academic research: what are the core concepts? What sources inform them? What questions remain open? What context matters?

The result is a vault I can actually navigate. Notes that connect to each other. A backlog of thinking I still want to do. Infrastructure instead of chaos. This a much bigger vault, so the structure expanded. Sources grew to books and authors, blog posts or frameworks. The fundamentals remained, allowing patterns to emerge organically.

Why the Same Structure Works Everywhere

Here’s what I didn’t expect when I built Laura: the basic architecture of knowledge is domain-independent.

Concepts - the building blocks of understanding. Whether you’re researching Victorian provenance or transit-oriented development, you need clear definitions of the ideas you’re working with.

Sources - where the knowledge comes from. Academic papers, auction catalogues, company reports, interview transcripts. Different materials, same need: structured notes with proper attribution.

Questions - what you don’t know yet. Every domain has gaps, contradictions, areas needing investigation. Making them explicit is half the battle.

Context - the situation you’re operating in. A dissertation has different constraints than job prep. The structure accommodates both.

This isn’t revolutionary insight. Researchers have known for centuries that good notes follow similar patterns regardless of subject. What’s new is having an AI that enforces the pattern consistently, creating files that persist and connect.

Faster, Cheaper, Better

Laura 1.0 worked, but she was expensive. Long sessions would hit token limits, forcing conversation compression that lost context. Complex vaults accumulated so much material that every interaction became slow and costly.

Rewriting the skill files to be leaner - more directive, less repetitive, trusting Claude to infer what didn’t need spelling out - reduced token usage by roughly 65%. The forced timeouts largely disappeared. Sessions that previously hit limits now completed comfortably.

This matters because research is iterative. You don’t build a knowledge base in one session. You return, add, refine, connect. If every session is expensive and fragile, the tool becomes impractical for real work. Making Laura efficient made her usable.

What I’ve Learned

Structure is transferable. The same information architecture works for art history, transport planning, job interviews, and personal knowledge management. Concepts, sources, questions, context. Universal primitives.

Persistence beats conversation. Chat is ephemeral. Files are infrastructure. The painting provenance research survived a six-month gap because it existed as artefacts, not memory.

Research is research. Academic investigation, professional preparation, personal curiosity - they’re all the same activity. Gathering information, structuring it, identifying gaps, building understanding. The domain changes; the methodology doesn’t.

Efficiency enables iteration. Reducing token usage wasn’t just cost optimisation. It made the difference between a tool you use once and a tool you live in.

The Unexpected Applications

I built Laura for academic research. She’s become general-purpose thinking infrastructure.

Next experiment: using her collaboratively with other personas. Laura researches; Alex (the solution architect) evaluates; Riley (the product owner) prioritises. Same vault, multiple perspectives. The painting provenance work could benefit from this - Laura gathering evidence, a detective persona testing theories, a writer persona drafting the narrative.

But that’s for another post.

---

The Obsidian Research Assistant is open source on GitHub. The basic structure - concepts, sources, questions, context - might be exactly what your next research project needs.