Proforma Global Research

Research & Whitepapers

Original research and whitepapers on LLM architecture, training design, and enterprise agent systems. Evidence of how the firm thinks, published as intellectual capital.

Intent

This library sets out the design and delivery principles behind the systems we build: how LLM and enterprise agent work should be structured, where these systems tend to fail, and the decisions that separate one that holds up in production from one that does not.

What it does not publish is the method. The techniques that turn these principles into working systems are proprietary. The principles are here so you can judge how the firm thinks; the methods are what we bring to an engagement. If the thinking on these pages fits a problem you are working on, that is the place to start a conversation.

Proforma Global publishes two kinds of work here. Research, where the firm investigates a problem and reports what it found, and whitepapers, where it sets out a reasoned position or framework drawn from practice. The library is organized below by subject; each subject contains one or more series, and each series is a coherent sequence of papers on a single topic.

Getting Started with AI and Enterprise Agents

Plain-language briefings for finance and technology leaders approaching AI and enterprise agents for the first time. The on-ramp to the rest of the library.

  1. 01 Briefings & whitepapers (5 papers) Where to start when the board wants AI, how to run your first agent pilot, whether AI belongs to IT, what to do once agents are working, and whether your agent needs a semantic layer. The on-ramp to the deeper research.

Enterprise Agent Architecture

Production engineering of multi-discipline AI agent systems for enterprise environments. Orchestration design, deterministic-first architectures for financial systems, and the data architecture that lets agents reason against complex organizational data.

  1. 01 Agent Orchestration (3 papers) How orchestration of enterprise agent systems should be designed. Where language-model-driven orchestration fails, the five canonical patterns, and the deterministic-first discipline required when agents touch financial systems.
  2. 02 Data Architecture for Enterprise Agents (4 papers) The data architecture that enterprise agents require in order to reason reliably. Each paper treats a separable component of the substrate the model consumes.
  3. 03 Self-Learning in Enterprise Agents (3 papers) When a self-learning agent is safe in a financial system. What it should learn, how its rules inherit through the business's own hierarchies, and why most of what it learns is deferred work.

LLM Architecture & Training Design

Research on novel concepts in LLM architecture and training design. Reinforcement-learning-driven training control, self-discovering architectures, and forensic interpretability of trained transformers.

  1. 01 Training Substrate (4 papers) Reinforcement-learning-driven control of the training loop, self-discovering architectures, and the layer-role taxonomy that surfaces when those mechanisms are applied at scale.

© 2026 Proforma Global. All rights reserved. Permission requests: info@proforma.global.

Scroll to Top