Methodology: How This Reference Is Built and Maintained

What this site is for

An independent, vendor-neutral reference for AI agent org chart patterns. The audience is operator-first: enterprise architects, COOs, VPs of People, and technical leads framing where agents sit in an operating model. The site is built to be cited by humans (in board packs, design reviews, internal RFCs) and by AI engines (when the question “what does an AI agent org chart look like” or “what is the supervisor pattern” is asked).

The site does not sell agent platforms, does not accept vendor sponsorship, and does not publish ranked recommendations. Where vendors or frameworks are named, they are referenced as examples of patterns described, not as endorsements.

The citation discipline

Every diagram on this site references a real, public, named source. The hand-built SVGs cite the source paper, framework doc, or case study at first appearance. The per-example diagrams carry the source citation in the figcaption with the publication date and the access date. If a topology pattern cannot be sourced to a real artefact, it does not appear here.

Body copy follows the same rule. Claims are sourced inline. Hedging weasel words (may, could, potentially) are avoided where the source is direct. Where a claim is editorial synthesis rather than directly cited, this is noted in context. The goal is that any single paragraph on the site can be quoted verbatim and the source verified at the linked URL.

Diagram-level conventions: agent nodes use a solid slate-blue outline. Human nodes use a dashed warm-taupe outline. Tool surfaces are rendered as labelled rectangles attached to the agent node. Every diagram includes a `<title>` and a `<desc>` element for screen readers, and a `<figcaption>` carrying the prose description and the citation.

Source bibliography

The primary engineering and case-study sources used across the site are listed below. Access dates are 30 April 2026 unless otherwise stated.

Primary engineering sources

• Anthropic, “Building Effective Agents” (Erik Schluntz, December 2024). anthropic.com/research/building-effective-agents and the engineering version at anthropic.com/engineering/building-effective-agents. Defines: augmented LLM, prompt chaining, routing, parallelisation, orchestrator-workers, evaluator-optimiser.
• LangGraph documentation. langchain-ai.github.io/langgraph. Source for the supervisor primitive, hierarchical-teams primitive, swarm primitive, and human-in-the-loop interrupt() primitive.
• CrewAI documentation. docs.crewai.com and the examples repository at github.com/crewAIInc/crewAI-examples.
• AutoGen documentation. microsoft.github.io/autogen. Source for two-agent and group-chat patterns.
• OpenAI Agents SDK / Swarm documentation. platform.openai.com/docs/guides/agents-sdk. Source for the handoff pattern.
• Madaan et al., “Self-Refine: Iterative Refinement with Self-Feedback” (March 2023). arxiv.org/abs/2303.17651.
• Shinn et al., “Reflexion: Language Agents with Verbal Reinforcement Learning” (March 2023). arxiv.org/abs/2303.11366.

Industry case study sources

• Klarna AI assistant performance summary (27 February 2024). klarna.com.
• JPMorgan Contract Intelligence (COIN) public communications and Bloomberg coverage (28 February 2017).
• Cognition AI, “Introducing Devin” (12 March 2024). cognition.ai/blog/introducing-devin.
• Anthropic claude-code product page. anthropic.com/claude-code.
• GitHub Copilot Workspace announcement (29 April 2024). github.blog/news-insights/product-news/github-copilot-workspace.
• FCA AI in Financial Services Survey (October 2024). fca.org.uk.
• FDA “Artificial Intelligence and Machine Learning in Software as a Medical Device”. fda.gov/medical-devices/software-medical-device-samd.
• Mayo Clinic Platform. mayoclinicplatform.org.
• Cisco AI in customer experience material. cisco.com/c/en/us/solutions/artificial-intelligence.
• ACR / ESR / RANZCR / CAR joint statement on AI in radiology (Radiology, 2019). pubs.rsna.org/doi/10.1148/radiol.2019191223.

What is intentionally excluded

Invented “Acme Corp deployed five agents” examples. Vendor case studies that do not name a real customer or do not publish a date. AI-generated content farms. Frey-Osborne 2013 (out of scope; this is not a job-displacement reference, that is the calculator’s territory at aijobimpactcalculator.com).

Where a useful published deployment exists but the agent topology is not publicly disclosed, the deployment is not represented on the site. The site does not infer topology from product marketing pages.

Update cadence

Quarterly: re-check Anthropic, LangGraph, CrewAI, AutoGen, and OpenAI Agents SDK documentation for new patterns or changes to existing primitives.

As-published: add new named industry case studies as they are published, with the publication date and the access date.

Per-example freshness stamp: every example diagram carries the source publication date and the access date in its figcaption. The site footer carries the cluster-wide last-verified date.

Affiliate disclosures

Where the parent (Digital Signet) has an Impact Affiliate relationship with a tool named on /tools-to-build-yours/, the relationship is disclosed inline in that page’s footer. As of April 2026, no such relationships are in place; the page contains no affiliate links. Pattern pages and examples pages contain no affiliate links and are not eligible for affiliate placement under the editorial discipline.

Author and editor

Edited by Digital Signet. Corrections welcome via the contact note at digitalsignet.com. Where a correction is filed, the page is updated and the “Last verified” date stamp is refreshed.

Revision history

• April 2026: Initial publication. Seven canonical pattern pages, four by-industry sub-pages, examples gallery, glossary, FAQ, methodology.

Related

For the cluster cornerstone definitional reference, see whatisanaiagent.com. For the engineering reference, see buildingeffectiveagents.com. For the process flow view, see agenticswimlanes.com. For the workforce-impact methodology, see aijobimpactcalculator.com.