AI tools for the building products industry, starting with automated order processing.
Category
Company
Published
Feb 12, 2025
Status
Active
Like any good product, BuildSuite was founded from the identification of one problem, there's too much manual entry in manufacturing.
Manufacturers rely on complex order intake processes, often receiving purchase orders via emails, PDFs, faxes, and handwritten forms. These unstructured formats require manual data entry, a tedious and error-prone process that slows down operations. The challenge was clear, how do we extract structured data from these unpredictable formats at scale?
Over a six-month development cycle, I led the team behind EntryIQ, an AI-powered solution that automates this process, reducing labor costs and improving efficiency. Designed to handle millions of order variations, EntryIQ converts unstructured order data into structured, export-ready formats, achieving up to 90% automation through its recursive AI prompt system.
At its core, EntryIQ is an intelligent pipeline for transforming raw, unstructured orders into structured, exportable data. The system automates order entry in a three-step workflow:
1. Order Ingestion and Preprocessing
Orders are forwarded into the system via email auto-forwarders or manual uploads. The system then classifies them based on format:
Plain text emails are sent directly to the AI extraction process.
Scanned PDFs or faxes undergo OCR (Optical Character Recognition) to extract readable text.
The goal of this step is to convert any order format into a text-based structure, ensuring it can be processed by the AI.
2. AI-Driven Data Extraction
Once an order is ready, it moves through EntryIQ’s recursive AI prompt system, which follows a hierarchical decision tree to extract relevant data.
Each node in the prompt tree represents a potential field (order number, item descriptions, shipping details, etc.).
The system evaluates each node dynamically, determining:
Whether the field exists in the document.
If present, what data should be extracted.
If absent, whether alternative pathways should be followed.
By breaking down the document step by step, EntryIQ can process orders without needing rigid templates. The recursive nature of the prompt tree allows the system to handle highly variable order structures, making it adaptable to new suppliers, changing formats, and unique order configurations.
3. User Verification and Export
Once the AI has extracted the necessary data, the order appears in the EntryIQ dashboard, where users can:
Assign themselves an order for review.
Compare AI-extracted fields to the original document.
Make corrections as needed before approval.
After verification, orders can be exported through four different methods:
Custom file formats (CSV, XML, Excel).
Direct API integrations with ERP or order management systems.
A quick copy-paste tool for manual entry when needed.
A default JSON export for interoperability with external systems.
This modular export system ensures that EntryIQ can integrate with any manufacturer’s existing workflow without requiring changes to their backend infrastructure.
EntryIQ’s AI performance is fully configurable by administrators, allowing them to fine-tune extractions without needing programming expertise.
The configuration page gives admins direct control over:
Custom prompt design – Defining how the AI should interpret and extract fields.
Real-time extraction previews – A sidebar shows how AI-extracted data would appear for manual verification.
Flexible prompt tree editing – Allowing continuous refinement as new order formats emerge.
By making AI customization accessible to non-technical users, EntryIQ enables rapid adaptation to evolving order formats, reducing reliance on software engineers to maintain automation rules.
To maintain high accuracy, EntryIQ includes Purkinje, a self-improving AI protocol designed to identify errors and refine extraction logic over time.
How Purkinje Works:
Throughout the day, EntryIQ logs all incorrect AI extractions, storing them in a training dataset.
The Purkinje protocol analyzes patterns in these errors, identifying consistent misclassifications.
The system automatically refines user-configured prompts, generating new extraction rules.
Before applying updates, Purkinje tests the improved prompts against a benchmark defined by administrators.
If the new configuration outperforms the original, EntryIQ automatically updates its AI logic, reducing future extraction errors.
Purkinje eliminates the need for constant manual tuning, ensuring EntryIQ adapts to variations over time.
Beyond automation, EntryIQ provides administrators with visibility into human verification efforts, ensuring that AI-driven processes remain accurate and reliable.
Employee performance tracking – The system records how often users correct AI-generated extractions, highlighting areas where further AI improvement is needed.
Correction trend analysis – Admins can identify common misclassifications and adjust AI training accordingly.
Benchmarking tools – Allowing businesses to validate AI performance before deploying updates at scale.
This oversight ensures that EntryIQ maintains a balance between automation and human validation, reducing errors while maximizing efficiency.
In Conclusion
EntryIQ has already been deployed in manufacturing environments, where it has delivered significant efficiency gains. A $100 million customer using EntryIQ has projected $500,000 in annual savings by reducing manual data entry costs and increasing order processing speed.
Key results:
Projected 90 percent automation achieved in structured order formats.
Reduction in manual processing time, enabling staff to focus on higher-value tasks.
Seamless integration with existing ERP systems, eliminating workflow disruptions.
Looking ahead, EntryIQ will continue expanding its capabilities:
Refining AI-driven prompt optimization for even greater accuracy.
Enhancing benchmarking tools to improve AI training efficiency.
Expanding ERP integrations to further streamline automation.
By combining recursive AI extraction, continuous self-improvement, and user-driven oversight, EntryIQ demonstrates how structured AI automation can transform order processing at scale.
Connect
Reach me at @chipherndon or cherndon@ucla.edu