top of page

Why Core Runway Adopted This Architecture in Release 4.71

Traditional search systems were built for a different era. They were designed to return links, rank pages based on keywords, and leave the human user to do the hard work of deciding what matters.

That model no longer fits modern business workflows.

When teams need accurate answers, structured data, or compliance-grade outputs, simply returning ten blue links is inefficient. What matters now is finding the right information quickly, evaluating it intelligently, and prioritizing what deserves attention.

That is why Core Runway has adopted a Scouts and Judges retrieval architecture in our latest Release 4.71.

The Problem with Conventional Search

Most internet retrieval systems still struggle with three persistent issues:

1. Too Much Noise

The web contains enormous volumes of duplicated, outdated, promotional, or low-quality content.

2. Weak Relevance Signals

Keyword matches do not necessarily mean business relevance. A page can mention the right terms while being practically useless.

3. Human Bottlenecks

Even after retrieval, someone still needs to inspect sources, compare them, verify freshness, and decide what to trust.

For enterprise workflows, that manual review becomes expensive and slow.

Enter the Scouts and Judges Model

The Scouts and Judges framework separates retrieval into two intelligent layers:

Scouts = Fast Explorers

Scouts are lightweight agents optimized for speed and coverage.

Their role is to:

  • Search across multiple public and private sources

  • Gather candidate pages, documents, APIs, databases, and feeds

  • Detect patterns and metadata

  • Cast a wide net at low cost

  • Continuously refresh available sources

Think of Scouts as a distributed research team rapidly scanning the landscape.

Judges = Deep Evaluators

Judges are higher-reasoning agents focused on quality, accuracy, and prioritization.

Their role is to:

  • Evaluate whether content truly answers the query

  • Score trustworthiness and freshness

  • Compare conflicting sources

  • Extract structured facts

  • Rank what matters most

  • Reject noise confidently

Think of Judges as senior analysts reviewing the scouts’ findings.

Why This Architecture Works

Instead of using expensive reasoning models on everything, Core Runway applies intelligence where it matters most.

Step 1: Scouts Find Options

Hundreds of possible sources can be surfaced quickly.

Step 2: Judges Review Only the Best Candidates

Rather than analyzing the entire internet, Judges focus on shortlisted results.

Step 3: Prioritized Outputs Are Delivered

Users receive ranked, relevant, usable information—not clutter.

This dramatically improves both cost efficiency and decision quality.

Why It Matters for Business Users

For industries like construction, environmental compliance, supply chain reporting, and regulated operations, the difference is substantial.

A user may ask:

  • Which regulation changed most recently?

  • Which material data sheet is current?

  • Which waste reporting template applies to this jurisdiction?

  • Which supplier document is missing?

  • Which sources are highest confidence?

A standard search engine gives links.

A Scouts and Judges system gives answers with priorities.

Core Runway Release 4.71

With Release 4.71, Core Runway has integrated this architecture into our latest retrieval workflows.

That means our platform can now:

  • Search broader data environments faster

  • Reduce irrelevant retrieval results

  • Prioritize high-confidence sources

  • Improve structured extraction pipelines

  • Lower token and compute costs by reserving deep reasoning for final-stage evaluation

  • Create stronger audit trails for enterprise decisions

This is particularly valuable in workflows where users need defensible outputs rather than generic search results.

The Hidden Advantage: Learning Over Time

Every judging decision creates metadata:

  • Which sources were trusted

  • Which patterns led to false positives

  • Which domains are consistently accurate

  • Which formats contain useful information

That creates a compounding system where retrieval improves continuously.

The architecture becomes smarter with use.

The Future of Search Is Not Search

The next generation of retrieval systems will not be about finding pages.

They will be about:

  • Finding truth

  • Ranking utility

  • Reducing noise

  • Explaining confidence

  • Delivering action-ready results

That requires multiple intelligent agents working in sequence—not one monolithic search step.

Final Thought

Scouts and Judges is more than a technical design pattern. It is a smarter economic model for AI retrieval.

Use lightweight systems for exploration.Use premium reasoning only where judgment is required.

That principle sits at the core of Core Runway Release 4.71—and it is a key step toward faster, more accurate enterprise intelligence systems.

 
 
 

One of the biggest barriers to adopting AI in compliance workflows isn’t capability — it’s trust.

Construction teams, environmental consultants, and compliance officers aren’t just asking:“Can AI extract the data?”

They’re asking:“Can I prove it’s right?”

Today, we’re introducing a major step forward.


🔍 Verifiable AI Extraction — Now Live in Core Runway

Core Runway now automatically marks up source PDFs after data extraction, showing:

  • Exactly where each data point came from

  • Which field in your report it populated

  • A clear, visual link between source → extraction → output

This turns AI from a black box into a fully auditable system.


🧠 How It Works

  1. Upload your documentsWaste dockets, invoices, material submittals, VOC certificates — any standard compliance input.

  2. Gemini AI extracts structured dataOur pipeline reads and interprets the document, identifying relevant fields (weights, material types, certifications, etc.).

  3. Automatic PDF markupCore Runway highlights the exact regions of the document used for each extracted value.

  4. Field-level traceabilityEach highlighted section is tied directly to a specific field in your LEED, CALGreen, or internal report.

  5. Human-in-the-loop approvalYour team reviews, verifies, and approves — in seconds, not hours.


⚡ Why This Matters

1. Instant Auditability

No more hunting through PDFs to verify entries.

Every number in your report is now traceable back to its source in one click.

2. Faster Human Review

Instead of re-reading entire documents, reviewers can focus only on:

  • Highlighted sections

  • Extracted values

  • Exceptions

This reduces review time by 90%+ per document.

3. Increased Confidence in AI

AI adoption often stalls because teams don’t trust automated outputs.

By showing exactly how each value was derived, Core Runway enables:

  • Confident approvals

  • Reduced rework

  • Clear accountability

4. Built for Compliance (Not Just Automation)

This isn’t generic document AI.

It’s purpose-built for:

  • LEED reporting

  • CALGreen compliance

  • Environmental and materials tracking

Where accuracy and auditability aren’t optional — they’re required.


🏗️ Real-World Example

A waste hauler uploads 150+ haul tickets for a project.

Instead of manually entering:

  • Material types

  • Diversion weights

  • Facility destinations

Core Runway:

  • Extracts all values automatically

  • Highlights each value directly on the ticket

  • Populates the LEED waste report

The reviewer:

  • Clicks through highlighted fields

  • Confirms accuracy

  • Approves

What used to take hours now takes minutes.


🔐 From AI Automation to AI Accountability

There’s a reason many AI pilots fail to deliver ROI:they stop at automation.

Core Runway goes further.

We combine:

  • AI extraction

  • Visual traceability

  • Human verification

This creates a system that is not just faster — but defensible.


🚀 What This Unlocks

  • Scalable compliance across hundreds of projects

  • Reduced reliance on manual data entry

  • Faster submissions with higher confidence

  • A clear audit trail for owners, regulators, and certifying bodies


Final Thought

AI shouldn’t replace human judgment — it should amplify it.

By making every extracted value transparent and verifiable,Core Runway ensures your team stays in control — while moving 10x faster.


 
 
 

As LEED continues to evolve, the shift from LEED v4.1 to LEED v5 represents more than just incremental updates—it reflects a fundamental change in how sustainability is measured, verified, and reported during construction.

For general contractors, sustainability consultants, and project teams, the biggest impact isn’t just what needs to be reported—but how much, how often, and how connected that reporting must be.

Below is a practical breakdown of how construction-phase reporting is changing—and what it means operationally.


1. From Static Submissions to Continuous Data Reporting

LEED 4.1

  • Reporting is largely snapshot-based

  • Teams compile documentation at key milestones or at project closeout

  • Heavy reliance on manual uploads (PDFs, spreadsheets, cut sheets)

  • Limited real-time visibility into compliance status

LEED 5

  • Moves toward continuous, lifecycle-based reporting

  • Greater emphasis on performance tracking over time, not just design intent

  • Encourages (and increasingly expects) digitally structured data

  • Alignment with platforms like Arc for ongoing performance monitoring

👉 Impact:Teams will need systems that track compliance as work happens, not just at submission deadlines.


2. Increased Depth in Material Reporting (EPDs, HPDs, VOCs)

LEED 4.1

  • Focus on collection of documentation

    • Environmental Product Declarations (EPDs)

    • Health Product Declarations (HPDs)

    • VOC data from product datasheets

  • Reporting is often binary: document present vs not

LEED 5

  • Shifts toward quantified impact of materials

  • Greater integration of embodied carbon (GWP) into reporting

  • More scrutiny on data quality and completeness

  • Less tolerance for missing or placeholder documentation

👉 Impact:Simply attaching datasheets is no longer enough—teams must extract, structure, and validate the underlying data.


3. Waste & Circularity: From Diversion Rates to Full Transparency

LEED 4.1

  • Construction waste reporting focuses on:

    • Total waste generated

    • Diversion rates (% recycled vs landfill)

  • Typically tracked via waste hauler reports and spreadsheets

LEED 5

  • Expands toward circular economy principles

  • Likely requirements include:

    • More granular waste categorization

    • Traceability of materials

    • Alignment with regulations like California SB 54

  • Increased need for audit-ready, verifiable data

👉 Impact:Manual tracking via spreadsheets becomes fragile—data lineage and auditability become critical.


4. Carbon Becomes Central (Not Optional)

LEED 4.1

  • Embodied carbon is introduced but not consistently central across all credits

  • Often treated as an additional analysis

LEED 5

  • Carbon is a primary metric across the system

  • Requires:

    • Integration of EPD data into carbon calculations

    • Clear reporting of Global Warming Potential (GWP)

  • Links construction decisions directly to climate outcomes

👉 Impact:Construction teams must connect procurement data to carbon reporting in near real-time.


5. Auditability and Data Integrity Requirements Increase

LEED 4.1

  • Review process relies on submitted documentation

  • Audit trails are often implicit or manual

LEED 5

  • Stronger emphasis on:

    • Traceability to source documents

    • Consistency across submissions

    • Reduced tolerance for errors or gaps

  • Aligns with broader ESG and regulatory reporting standards (ISSB, CSRD)

👉 Impact:Every reported number must be defensible, traceable, and reproducible.

6. Integration with Construction Workflows

LEED 4.1

  • Reporting is often parallel to construction workflows

  • Data is re-entered from systems like Procore or spreadsheets

LEED 5

  • Push toward embedded reporting within project systems

  • Integration with:

    • Procore

    • Autodesk

    • Procurement and material tracking tools

👉 Impact:The future is “reporting without reporting”—data captured once, used everywhere.


What This Means for Project Teams

Across all categories, the direction is clear:

Shift

LEED 4.1

LEED 5

Reporting Style

Periodic

Continuous

Data Format

Documents

Structured data

Carbon

Emerging

Core metric

Waste

Summary-based

Detailed + traceable

Auditability

Manual

System-driven

Workflow Integration

Limited

Required

The Bottom Line

LEED 5 doesn’t just increase reporting requirements—it changes the operating model.

  • Manual workflows → Automated data pipelines

  • Document collection → Data extraction and validation

  • End-of-project reporting → Real-time compliance tracking

For teams still relying on spreadsheets and manual uploads, the gap between current workflows and future requirements is significant.

Those who adapt early—by digitizing data flows and embedding compliance into construction workflows—will not only reduce reporting burden but gain a competitive edge in delivering faster, more predictable, and audit-ready projects.


Looking Ahead

As LEED 5 rolls out, expect further alignment with:

  • State-level regulations (e.g., CALGreen 2026, LL97)

  • ESG disclosure frameworks

  • Real-time performance platforms like Arc

The winners won’t be those who report better—they’ll be those who design systems where reporting happens automatically.

 
 
 
bottom of page