Skip to content

Explaining FCC to the Public

Good AI coverage requires short, accurate, memorable explanations. This page gives you reusable analogies, tested opening lines, and the two or three concepts you should never omit.

The three concepts every public-facing piece needs

  1. Multi-agent means multi-step. FCC does its work in discrete phases rather than one big call.
  2. Roles are documented. Each agent role is defined in a structured specification (R.I.S.C.E.A.R.). Readers do not need to memorize the acronym; they need to know that documentation exists.
  3. Outputs are auditable. Every run produces a machine-readable trace. Regulators and journalists can inspect it.

If a reader takes only those three ideas away, your coverage is accurate enough.

Tested analogies (use at most one)

  • The editorial newsroom. "One staffer researches, another drafts, a third edits with authority to kill the story. FCC systems are built along this separation of duties."
  • The restaurant kitchen. "The prep cook readies ingredients, the line cook plates, the expediter checks before the plate leaves the kitchen."
  • The courtroom. "Advocate presents, judge assesses, clerk records. The separation is not decoration; it is how accountability works."
  • The peer-review cycle. "A researcher drafts, a referee critiques, and the journal records the exchange. Each role is defined in a published style guide."

Pick one and stick to it through the piece. Mixing analogies confuses readers.

Opening paragraphs you can adapt

Lead for a feature (public-sector deployment story):

"When the Springfield municipal records office installed its new AI triage system last April, the city's chief counsel insisted on seeing one document before signing the procurement: the 'ComplianceReport.' The report, produced automatically by a framework called FCC, listed every regulatory requirement the system was designed to satisfy, and which of 256 European AI Act provisions the deployment had tested. The counsel read it in an afternoon. That such a document exists at all is unusual; most AI procurements ship with marketing."

Lead for a news piece (research announcement):

"A research group today published a study using FCC, an open-source framework that structures AI agents into roles and logs every step of their reasoning. The framework, maintained by INFORMATION COLLECTIVE, LLC, is unusual in that it produces auditable traces and risk-classified documentation for every run. The study claims a measurable reduction in error rates across ten reasoning tasks."

Lead for a critical piece:

"The vendor's claim of '100% compliance' rests on a document it would not share. The document exists in principle — every FCC-based system produces one — and independent reviewers have been requesting it for three weeks without result. The pattern, advocates say, is the reason open frameworks like FCC were invented."

Each lead embeds the key concept (auditable, structured, open) without lecturing.

Words to avoid

  • "Sentient." FCC systems are not sentient. They are not even models; they are frameworks around models.
  • "Neural network." FCC is not one. It may call one, but that is incidental.
  • "Artificial intelligence" as a mass noun, standing alone. Prefer "the deployment," "the system," "the framework."
  • "Revolutionary," "game-changer," "paradigm shift." These words are PR, not reporting.
  • "Hallucinate" — technically imprecise; prefer "fabricate" or "generate unsupported content."

Words to use

  • Framework. What FCC is.
  • Deployment. A specific instance of a framework in use.
  • Persona. A documented role in the deployment.
  • Workflow. The sequence of steps from input to output.
  • Audit trail / trace. The machine-readable log of what happened.
  • Risk classification. The EU AI Act categorization of a component.

Length-specific templates

100 words (brief). Name the deployment, its domain, its risk category, the vendor, the framework version, the immediate news. Example: "Springfield's municipal records office has gone live with an FCC-based triage system built on the open-source Crater constellation-vertical. The deployment carries a high-risk classification under the EU AI Act, with 211 of 256 applicable requirements fully compliant on the pre-launch audit. Closure of two outstanding findings is required by June 1 under the city's conditional approval."

300 words (news). The above plus: one sentence on how FCC works; one sentence of critic or context; one sentence on next steps. Aim for two named sources.

800 words (feature). Add background on FCC, a paragraph on a sister project, a paragraph on reader impact, a paragraph on open questions. Three or four named sources.

2,000+ words (long-form). Add a worked example walked through by a principal engineer, a discussion of threats to validity, a contrast with a competing architecture, and a short methodology box describing your reproducibility work.

Common misreadings to prevent

Readers often misunderstand these points. Inoculate them:

  • Persona count. A "164-persona framework" does not mean any deployment uses 164. Most use 3-10.
  • Risk category. Not a quality score. "High-risk" means "regulated," not "bad."
  • Open source. Means the framework is inspectable. Does not mean the deployment is free or audited.
  • Mock mode. Means "no LLM called." Safe for demos; not equivalent to production.

Reader comprehension test

Before filing, ask a colleague who does not cover AI to read the piece and tell you back what they learned. If they say "I understand the system has stages and someone can check each stage," you have succeeded. If they say "sounds cool," you have not.

Attribution micro-policy

When you rely on an FCC artifact in a story, mention it by name once. Examples:

  • "According to the deployment's ComplianceReport, dated April 23, 2026..."
  • "The system's Critique persona is documented to have halt authority..."
  • "The research group shared their scenario files and FCC v1.7.0 pin..."

Named-artifact citations are more credible than vague references to "the system."

Where next

With the public-explainer in hand, sharpen your citations at Citation Guide, or prepare interviews with Interview Personas. For sources beyond FCC, Further Reading.