ILR 2.0 — Innovation Lineage Record

Inspiration

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for ILR 2.0 — Innovation Lineage Record

Inspiration

Important work is often preserved as a final document, meeting summary, or AI answer. The path behind it becomes fragmented across notes and conversations. That makes it difficult to reconstruct where an idea began, why a decision changed, what evidence mattered, or which uncertainties remain.

ILR began with a simple question: Can you explain not only what you know, but show the complete path by which you came to know it?

What it does

ILR creates one connected record from the first curiosity to the final conclusion.

Every meaningful entry receives a timestamp and a distinct role such as question, observation, evidence, hypothesis, revision, or conclusion. Chronos preserves the sequence. Kairos identifies moments that redirect understanding. The Knowledge Growth Matrix makes relationships visible. A conclusion can be traced backward to the entries supporting it.

GPT-5.6 assists by suggesting classifications, evidence relationships, possible turning points, and provisional conclusions. These suggestions remain visibly separate from human-authored content. The researcher decides what to accept, revise, or reject.

How we built it

The prototype was built with Codex as the primary product and engineering collaborator. Codex helped translate the ILR constitutional framework into a focused seven-screen MVP, define the data model, implement persistence and OpenAI integration, design the evidence-lineage views, deploy the application, and simplify the final judge experience.

GPT-5.6 is integrated through the Responses API using Structured Outputs. Luna handles frequent entry analysis efficiently, while Terra handles the more demanding provisional conclusion. Their role is constrained: organize the supplied record, never invent evidence, never decide truth, and never replace the human conclusion.

Challenges

The hardest challenge was not technical—it was preserving ILR's central philosophy while building software. Early versions made AI controls too prominent and risked turning ILR into another complicated assistant. We corrected that by returning human authorship to the center, collapsing model details behind review controls, and creating a guided demonstration that explains the framework before exposing advanced behavior.

We also separated the core framework from its applications. The guided scenario traces the live conference presentation, the 20–50-person Knowledge Growth Matrix, and the interactive judge experience into the working product. It demonstrates how ILR preserves relationships, dates, differences, and uncertainty without losing the creator's original thought process.

Accomplishments

• A working seven-screen product
• A complete curiosity-to-conclusion lineage
• Chronos, Kairos, evidence lineage, and Knowledge Growth views
• Durable session storage
• Human-editable classifications and conclusions
• GPT-5.6 structured analysis with visible human review
• A portable ILR export
• A guided judge experience
• A real idea-lineage demonstration built from ILR's presentation notes and dated records
• Expandable evidence-basis, source, and limitation explanations

What we learned

Traceability is not created merely by saving more text. It requires preserving distinctions: evidence versus interpretation, human authorship versus model suggestion, chronological order versus pivotal significance, and a draft conclusion versus a confirmed one.

We also learned that restraint is a feature. The strongest version of ILR is not the version with the most AI. It is the version where AI reduces organizational burden while human judgment remains unmistakable.

What's next

The current MVP deliberately supports one person and one investigation. Future work may add multi-user collaboration, source verification, stronger graph exploration, and live conference participation. Those are future directions; the Build Week prototype demonstrates one complete lineage clearly and reliably.

Built during Build Week

Pre-existing work: ILR's conceptual framework, including Chronos, Kairos, evidence lineage, and its founding philosophy.

Meaningfully extended during the submission period: the complete software product, seven-screen interaction model, structured data flow, persistence, GPT-5.6 integration, guided demo, export, deployment, testing path, and submission materials.

Portable ILR — phone access

Scan the Portable ILR QR code on the public GitHub project page, or open research mode directly. This phone entry point starts the same timestamped research workflow while preserving the stable prototype and the safeguarded portable standard.

Built With

• openai-responses-api