Customers,
past & current.

Use case

Pump health monitoring, predictive maintenance, and operational reporting across produced water infrastructure: pumps, pipelines, storage, and treatment facilities. SWD facility monitoring across 7 sites, with pump performance modeled via a custom ML transform.

• Sites: Sheepshead, University U2, University 21-1, Espinoza 10, Hartgrove, Stiles, Gunter
• 3 sites have kWh metering, 4 don't
• Custom pump_metrics_transform model handles Hz scaling per site group
• Field Assist (conversational AI), a mobile platform, and a digitized expenditure approval workflow round out the deployment

Components used

Use case

Turning high-frequency sensor data from fracturing operations into clear operational insight: stage transitions, equipment behavior, and pumping efficiency. Sample notebook + ds guide in docs/.

Components used

Use case

Cleaning up fragmented inventory and equipment data: duplicate materials, non-moving inventory, and equipment nearing obsolescence.

• Live platform at freeport.neuralix.ai
• Covers material and equipment portfolios
• Duplicate detection, usage forecasting, and stocking recommendations
• Automated ingestion from existing enterprise systems

In contract negotiations toward full production.

Components used

Use case

Automating the mapping of Completions data into industry-standard reporting formats, replacing a manual process running at roughly 70% accuracy.

• Built on Neuralix's NIX platform
• Automated data mapping
• Real-time validation
• Performance benchmarking dashboards

Status

Pilot. Development began December 2024.

Use case

This is Phase 2 of the customer's supply-chain AI enablement program. Phase 1 (built independently of Neuralix) proved that AI-driven classification of supply-chain spend was achievable, reaching high-90s confidence on primary category classification, but left the underlying data foundation incomplete: ingestion was a frozen snapshot rather than live, units of measure weren't standardized, and there was no governed way for the business to access the curated data directly.

Phase 2 is foundation-first: bring in the operational data (well activity, production volumes, scheduling) that gives spend numbers their context, so cost can be normalized per unit of actual output rather than judged on rate sheets alone. On top of that foundation, the proposal adds productized category-decision dashboards, a small set of specialist AI agents for high-spend procurement categories, an orchestration layer so those agents can be queried from one place, and a human-in-the-loop review workflow so low-confidence outputs route to a person rather than silently failing.

Tech & status

Proposed to run on DLT, replatforming the customer's existing pipeline onto Neuralix's own data-lifecycle templates rather than continuing on their prior stack. The engagement is scoped as a staged, multi-month delivery: a discovery stage, a data-foundation stage, a dashboard/platform stage, and an agents stage, with an initial governed, queryable layer over existing spend data targeted within the first 30 days of signature.

The proposal has been submitted and evaluated. Per Abhishek: close to a signed contract, but not yet formally under contract as of this writing.

Use case

A fleet decarbonization challenge: optimizing vehicle routing to move goods while minimizing emissions. Entered as a startup hackathon, which Neuralix won.

Tech & status

Pure data science modeling, no platform involved. Scoped and closed at the hackathon itself, this was not a pilot or ongoing engagement on its own.

Use case

The hackathon win led to a separate, follow-on project with Shell. This is a distinct engagement from the hackathon itself, not a continuation of the same scope.

Tech & status

Use case

Predictive quality assurance on layered substrate manufacturing used in RNA and DNA testing. Their existing process took two full days after production to catch a defect. Related to Agilent's internal product line for this work (exact product name to be confirmed against Agilent's public materials before publishing).

Tech & status

Time-series processing and modeling on sensor data (temperature, pressure, print-head diameter) to predict quality issues up to two days earlier than the existing process. DLT was proposed as part of the solution but never reached a live demo. Closed after Agilent went through a company reorganization and the project was deprioritized.

Use case

Converting unstructured, Word-based pharmaceutical manufacturing "recipe" documents into structured data for import into a factory automation system, reducing manual, error-prone re-entry across multiple factories.

Tech & status

An AI agent that reads the document, identifies its structure, and outputs structured data automatically, Neuralix's first agentic AI project. Closed when Deloitte's primary contact for the project moved on.

Use case

Financial analysis on a loan portfolio issued largely without standard documentation or credit checks. A separate, broader proposal also existed to evaluate their AI tools and rebuild core business processes around AI.

Tech & status

Financial and data modeling on the loan portfolio, no platform deployment. Closed: the client went bankrupt while the engagement was still active.