Crivella is a Pittsburgh systems-engineering firm, founded in May 1986 by Arthur R. Crivella and Wayne J. West. For four decades it has worked the same problem from different angles: how organizations capture, structure, and operationalize the judgment of their most experienced people. Two granted U.S. patents — US 6,883,008 (2001) and US 7,779,007 (2007) — describe what the AI industry now calls retrieval-augmented generation with provenance, two decades before it had a name.
Since 1986
Founded May 1986 · Founder-Present Today
2 U.S. Patents
US 6,883,008 (2001) & US 7,779,007 (2007)
~20 People
A Pittsburgh Boutique
Arthur R. Crivella and Wayne J. West founded the firm in May 1986. They worked together for the next three decades, co-inventing the two patents the platform still rests on, until Wayne's death in 2019. The firm operates today with younger operating leadership; Arthur continues as Inventor, shareholder, and engineer.
Arthur's background is in economics and mathematics, and he has been building large-scale computer systems since 1978. He is the firm's founder, inventor, and architect — not a marketing figure and not a manager who never wrote code. The technical work is his own.
Complex problems — whether industrial or legal — are fundamentally systems and information problems.
Crivella
The forty years are not a slogan. They are a corporate record. The same engineering practice, the same founder, the same patents — under successive entities, operated today by Tera Alta Risk, Inc.
Advanced Systems Engineering, Inc.
Founded in May by Arthur R. Crivella and Wayne J. West.
ASE Edge, Inc.
The practice carried forward as the work moved into knowledge systems.
Arete Legal, Inc.
The litigation era — the work that became 80+ federal MDLs.
Tera Alta Risk, Inc.
Operates the Crivella brand today, holding the same patents and practice.
The Crivella platform does not trace its origins to the current wave of AI enthusiasm. It traces them to two inventions, separated by years, that together constitute the complete architecture of what the AI industry now calls enterprise knowledge management and retrieval-augmented generation.
The first, US 6,883,008 (filed 2001), created the knowledge repository: a system combining library science, linguistics, and computer technology to ingest, organize, and make accessible large volumes of documents and data — with multi-level privileged access, advanced search, and a schema purpose-built for multi-district litigation.
The second, US 7,779,007 (filed 2007), created the content identification engine: a method for analyzing a corpus, applying marker sets to identify relevant content, scoring each document against a reference threshold, and iteratively refining the process until the output is a precisely bounded collection — containing exactly what is relevant to a given task, and nothing that is not. This is what produces Context Collections. This is what makes the AI outputs accurate.
This architecture is also the structural answer to what the AI industry now calls hallucination. AI that reasons from training data produces confident outputs that cannot be traced to a source. AI that reasons from a curated, reliability-graded evidence collection — grounded AI — produces outputs that are traceable, verifiable, and defensible. The Crivella platform was designed to produce grounded AI output from the beginning. That was the point of the architecture in 2001. It remains the point today.
2001
The Knowledge Repository · US 6,883,008
2007
The Content Identification Engine · US 7,779,007
The architecture the AI industry calls retrieval-augmented generation — a managed knowledge repository feeding precisely curated context to AI systems — was co-invented by Arthur R. Crivella and Wayne J. West in 2001 and 2007. Before transformers. Before GPT. Before the current AI wave began.
Every concept in today's platform has been tested and improved in real-world deployment since 2001 — across the most adversarial information environments in American practice. What the industry is building now, Crivella has been building and applying for decades.
The firm's work runs where the cost of being wrong is high — a sainthood inquiry, a national battlefield, a defense supply chain.
Newman Canonization
A global research platform for the Roman Catholic Church, supporting the canonization of John Henry Newman.
Gettysburg Museum
Knowledge systems for the National Park Service's Gettysburg National Battlefield Museum.
Marine Corps Depots
Depot modernization for the Department of Defense, preserving critical maintenance capability under BRAC.
80+ Federal MDLs
Litigation infrastructure across 80+ multi-district litigations representing 400,000+ claimants, under federal protective order.
The institutional work demands accuracy under scrutiny. The litigation work demands it under an adversary actively trying to keep evidence out of view. Both are the same discipline applied at the furthest frontier.
Crivella is a roughly 20-person boutique, not a venture-cycle startup chasing the AI news cycle. The firm is led day to day by a younger operating team. Arthur R. Crivella continues as the firm's Inventor, shareholder, and engineer — present, not retired.
That continuity is the point. Family is one of the firm's four founding values, alongside excellence in engineering, creativity, and operating at the furthest frontier of innovation. They were stated by the founders in 1986 and they still govern the work. Based in Pittsburgh, working everywhere.
How the Platform Works →Excellence in engineering
The technical quality of every output is non-negotiable. Specifics earn trust; adjectives don't.
Creativity
Solving problems no template covers — and naming the categories rather than borrowing the industry's vocabulary.
The furthest frontier
Forty years at the edge. The 2001 and 2007 patents anticipated grounded AI by two decades.
Family
A multi-generational firm, founder-present — institutional continuity over a venture cycle.
Tell us about your organization and where you want grounded AI to take you. We'll tell you what it actually takes to get there.