# GoPX.ai

> GoPX lifts documents into structured logic, binds AI outputs to verified rules, and makes every decision auditable. Model-agnostic. 85% peer-reviewed reduction in hallucination.

## About
GoPX.ai — Verified decision architecture for enterprise AI.
Website: https://gopx.ai

## What GoPX Does
Most enterprise AI is confident and wrong. GoPX lifts your documents into structured logic, binds every output to verified rules, and makes every decision auditable.

## Solution Surfaces
- **Legal AI**: Win more cases with AI that proves its reasoning. Run your case files through an AI that cites every precedent, flags every vulnerability, and ships an audit trail your court would accept. Built for litigation firms and in-house legal teams.
- **Healthcare**: Private clinical documentation that keeps clinicians in control. Turn private transcripts and patient context into editable SOAP notes, missing-fact checks, validation warnings, and follow-up questions. Designed for local deployment and clinician approval.
- **Agentic Commerce**: Verified decisions for autonomous commerce. When AI agents discover, compare, and transact at scale, hallucinated recommendations are operational risk. GoPX binds agent outputs to verified rules and source evidence before commerce networks let agents act.
- **Defense**: Sealed explanations for contested environments. Analysts and commanders need conclusions that survive adversarial scrutiny. GoPX seals each output to its evidence, doctrine, and rules — a tamper-evident record that can be audited, challenged, and reproduced.

## How It Works
- **Documents**: Start with the source material your team already trusts: policies, contracts, reports, transcripts, regulations, and operating playbooks.
- **Structured Logic**: Convert source text into explicit facts, relationships, rules, and exceptions that humans can inspect and systems can reuse.
- **Verified Rules**: Bind model outputs to tested constraints and cited evidence, so every decision is grounded before it reaches the user.
- **Auditable Decisions**: Produce final outputs with a traceable path from source material to rule logic to the decision your team has to defend.

## Proof Points
- **100%**: hard-constraint satisfaction across 180 production queries
- **113/113**: unit tests passing on the deployed reference stack
- **85%**: peer-reviewed reduction in hallucination and toxicity
- **30+ yrs**: of published research behind the approach
Peer-reviewed in IEEE Transactions on Computational Social Systems: "Beyond the Black Box: Programmable AI and Explainable Text Analysis." — Trivedi, Çetinkaya, Cowan, Newson, Vlahović, Davulcu (forthcoming).

## Decision Control Framework
Decision Control is the GoPX framework for what makes an AI system defensible in regulated use. Each question maps to a real failure mode of unrestricted LLMs — and to the specific GoPX capability that closes it.
- **Catastrophic forgetting**: GoPX doesn't fine-tune the model — it lifts your domain logic into a separate, inspectable rule layer. The base model stays general; the domain stays editable. Forgetting is not a failure mode because the knowledge was never in the weights.
- **Weak learning**: Lifting forces the signal into explicit rules. If the model is winning by exploiting a shortcut, the rule layer makes the shortcut visible — operators see the rule firing, audit it, and replace it with the actual signal.
- **Prompt injection**: The rule layer is not a prompt. Adversarial text can change what the model says next; it cannot change what the rule layer requires the model to cite. Lowering binds outputs to the rule layer, so injection attempts surface as constraint violations rather than silent rewrites.
- **Execution exposure**: Every action runs against the verified rule set with a recorded reasoning trace. Operators can require dual-control, escalation thresholds, and pre-execution review per action class — declared as logic, not buried in prompt strings.
- **Information sovereignty**: The rule layer and the audit trail run on infrastructure you choose. The model can be local, hosted, or proxied. The decisions and their evidence stay where you put them — no telemetry hostage.
- **LLM independence**: The rule layer is model-agnostic by construction. Switch from GPT-5 to Claude 4 to Gemini 3 to Llama-derived models without rewriting business logic. The cost of switching becomes evaluation cost, not rebuild cost.

## Founder
**Dr. Hasan Davulcu** — Founder, GoPX.ai · Professor, School of Computing and Augmented Intelligence, Arizona State University

Dr. Hasan Davulcu is a professor at Arizona State University's School of Computing and Augmented Intelligence, where he directs the Cognitive Information Processing Systems (CIPS-AI) Lab. For over two decades he has developed patented AI and machine learning algorithms for mining multi-lingual text, image, and video from social networks, and pioneered logic-based foundations for programmable agentic workflows. His work on sociocultural modeling earned the 2011 HSCB Focus Exceptional Scientific Achievement Award from the US Office of the Assistant Secretary of Defense for Research and Engineering.

### Credentials
- Ph.D. Computer Science, Stony Brook University
- M.S. Computer Science, Stony Brook University
- B.S. Mathematics, Middle East Technical University (METU)

### Profiles
- ASU Faculty Profile: https://faculty.engineering.asu.edu/davulcu
- Google Scholar: https://scholar.google.com/citations?user=P9IivpwAAAAJ&hl=en
- CIPS-AI Lab: https://faculty.engineering.asu.edu/davulcu

## Research
The intellectual foundations of GoPX.ai — peer-reviewed work on programmable AI, agentic workflows, and explainable reasoning.

### Featured: Beyond the Black Box: Programmable AI and Explainable Text Analysis for Trustworthy Social Intelligence
Trivedi, A., Çetinkaya, Y.M., Cowan, M.A., Newson, M., Vlahović, N., Davulcu, H.. *IEEE Transactions on Computational Social Systems (TCSS), Special Issue on Revolutionizing Social Intelligence with AI Technologies and Sensing Innovations*, Forthcoming.
A neuro-symbolic Programmable AI algorithm that turns neural outputs into interpretable, logic-patterned narratives via symbolic pattern lifting and lowering. Built for social scientists, it enables reproducible Human + AI text analysis across social media, interviews, and news — demonstrated on 1.3 million tweets about global energy trends. The paradigm behind GoPX: AI that not only says what it does, but does what it says.

### Selected publications
- Çetinkaya, Y.M., Lee, Y., Külah, E., Toroslu, I.H., Cowan, M.A., Davulcu, H., 2024. "Toward a Programmable Humanizing Artificial Intelligence Through Scalable Stance-Directed Architecture." *IEEE Internet Computing, Vol. 28(5), pp. 20–27 · Special Issue on Civilizing and Humanizing AI*.
- Çetinkaya, Y.M., Trivedi, A., Yanamandala, V.D., Cowan, M.A., Toroslu, I.H., Davulcu, H., 2025. "NARRA-SCALE: Scaling Users and Messaging Through Narrative Detection in Retweet Networks." *IEEE ICTAI 2025, pp. 187–194*.
- Çetinkaya, Y.M., Külah, E., Toroslu, I.H., Davulcu, H., 2024. "Targeted marketing on social media: utilizing text analysis to create personalized landing pages." *Social Network Analysis and Mining, Vol. 14(1), Article 77*.
- Mousavi, M., Davulcu, H., Ahmadi, M., Axelrod, R., Davis, R., Atran, S., 2022. "Effective Messaging on Social Media: What Makes Online Content Go Viral?." *The Web Conference (WWW 2022), pp. 2957–2966*.
- Davulcu, H., Kifer, M., Ramakrishnan, I.V., 2004. "CTR-S: A Logic for Specifying Contracts in Semantic Web Services." *WWW 2004*.
- Davulcu, H., Kifer, M., Ramakrishnan, C.R., Ramakrishnan, I.V., 1998. "Logic Based Modeling and Analysis of Workflows." *ACM PODS 1998*.

## FAQ
### What is GoPX.ai?
GoPX.ai builds verified decision architecture for enterprise AI. It lifts contracts, policies, regulations, catalogs, and other enterprise documents into structured logic, then binds model outputs to verified rules and source evidence.

### How is GoPX different from ChatGPT, LLaMA, or other LLMs?
GoPX does not try to make one model memorize your domain. It keeps domain logic outside the model in an inspectable rule layer, so teams can audit decisions, edit constraints, and switch models without rebuilding the business logic.

### Which enterprise AI risks does GoPX address?
GoPX is designed for the risks regulated buyers ask about first: catastrophic forgetting, weak learning, prompt injection, execution exposure, information sovereignty, and LLM independence.

### How does GoPX reduce inaccurate AI outputs?
GoPX constrains outputs to verified rules and indexed evidence. Every decision returns the clauses, policies, rules, and reasoning artifacts used to reach it, which makes failures easier to detect, correct, and prevent from recurring.

### Who is behind GoPX.ai?
GoPX is founded by Dr. Hasan Davulcu, Professor at Arizona State University's School of Computing and Augmented Intelligence and director of the CIPS-AI Lab. He has over two decades of peer-reviewed research on programmable AI, agentic workflows, sociocultural modeling, and explainable reasoning — published at ACM, IEEE, and NSF/DoD-funded venues.

### How do I learn more or get in touch?
The featured paper — 'Beyond the Black Box: Programmable AI and Explainable Text Analysis for Trustworthy Social Intelligence' — is the newest formal statement of the lifting/lowering approach and is forthcoming in IEEE Transactions on Computational Social Systems. For partnership and enterprise pilot inquiries, use the contact form below.

## Links
- Google Scholar: https://scholar.google.com/citations?user=P9IivpwAAAAJ&hl=en
- ASU Faculty: https://faculty.engineering.asu.edu/davulcu
- CIPS-AI Lab: https://faculty.engineering.asu.edu/davulcu