A single source of truth for personal brand

PersonSpec is an open standard for defining a person's professional voice and presence as structured, machine-readable JSON. Identity, why, pillars, voice, vocabulary, structure, and guardrails — all in one canonical file, aligned to the organisations a person represents.

A PersonSpec file is a single .json document. Define it once; consume it everywhere a leader's voice needs to land consistently.

1. Define the person

   Create a personspec.json file using the open schema. Capture identity, why, pillars, voice, vocabulary, structure, and guardrails. Reference one or more parent BrandSpecs to articulate how the person aligns to the organisations they represent.

2. Calibrate against a parent

   Each parent reference carries an alignment object — voice deltas, topic scope, and acknowledged tensions. PersonSpec makes the personal-to-organisational relationship explicit rather than assumed.

3. Push to your tools

   Feed the file into a content tool, ghostwriter brief, or AI assistant. The model receives a structured, unambiguous voice brief — and produces output that is calibrated, not generic.

4. Keep it live

   When a position evolves, a pillar shifts, or a new parent relationship begins, update the personspec.json. Every connected tool stays in sync. Voice drift becomes a configuration problem, not a communications problem.

A PersonSpec document is a JSON object with a root person key and seven top-level sections. All sections are optional at the object level but recommended for complete personal-brand definition. Identity is the only required field.

Voice fields

Alignment fields

The alignment object is what makes PersonSpec relational. Each parent reference carries one of these objects, articulating the relationship between the person and that organisation.

Provenance fields

PersonSpec describes a person — and the same shape can be authored by the subject themselves or inferred about them by a third party. Provenance makes that distinction explicit, so consumers can treat the two cases differently. The provenance block lives on metadata.provenance.

A condensed but complete PersonSpec document showing the multi-parent case. All fields shown are part of the v0.1.0 schema.

{
  "_export": {
    "personName":        "Jane Doe",
    "personSpecVersion": "0.1.0",
    "exportedAt":        "2026-05-02T10:00:00.000Z"
  },

  "person": {

    /* ── Identity ────────────────────────────── */
    "identity": {
      "name":     "Jane Doe",
      "role":     "Founder & Director",
      "headline": "Building purposeful brands at human scale.",
      "parents": [
        {
          "brandspec": "https://example.com/brandspec.json",
          "role":      "Founder & Director",
          "scope":     "primary",
          "alignment": {
            "voiceDelta": "warmer, more first-person",
            "topicScope": "all"
          }
        }
      ]
    },

    /* ── Why ─────────────────────────────────── */
    "why": {
      "archetype": "creator",
      "innerDevelopment": {
        "being":         4,
        "thinking":      5,
        "relating":      3,
        "collaborating": 3,
        "acting":        4
      },
      "statement": "I help small organisations make their brand work harder."
    },

    /* ── Pillars ─────────────────────────────── */
    "pillars": [
      {
        "name":         "Purposeful branding",
        "parentPillar": "purpose",
        "relationship": "reinforces"
      },
      {
        "name":         "Open standards",
        "parentPillar": "practice",
        "relationship": "extends"
      }
    ],

    /* ── Voice ───────────────────────────────── */
    "voice": {
      "tone":          ["curious", "warm", "direct"],
      "sentenceStyle": "mixed",
      "formality":     3,
      "humour":        "self-deprecating",
      "perspective":   "first-person"
    },

    /* ── Vocabulary ──────────────────────────── */
    "vocabulary": {
      "signaturePhrases": ["make it matter"],
      "avoid":            ["unlock", "leverage", "synergy"],
      "industryLanguage": ["design tokens", "open standard"]
    },

    /* ── Structure ───────────────────────────── */
    "structure": {
      "hookStyle":     "story",
      "defaultLength": "short",
      "signoffStyle":  "none"
    },

    /* ── Guardrails ──────────────────────────── */
    "guardrails": {
      "topicsAvoid":    ["party politics", "religion"],
      "positionsTaken": [
        "Open standards beat proprietary formats for SMEs."
      ],
      "confidentiality": "Never name unannounced clients or unreleased work."
    },

    /* ── Metadata ────────────────────────────── */
    "metadata": {
      "name":        "Jane Doe",
      "version":     "1.0.0",
      "lastUpdated": "2026-05-02T00:00:00.000Z",
      "provenance": {
        "type":       "self-reported",
        "consent":    "subject",
        "confidence": "high"
      }
    }

  },

  "$schema":  "https://personspec.org/schema/v0.1/personspec.json",
  "version":  "0.1.0"
}

PersonSpec is designed to be consumed by any tool that needs a structured definition of a person's voice and presence. Below are reference patterns for the most common outputs.

AI system prompt injection

Pass the entire person object as structured context in an AI system prompt. The model receives a complete operating brief — voice, vocabulary, structure, guardrails — and produces output calibrated to a specific person, not a generic tone.

// Build a calibrated AI brief from PersonSpec
const spec = await fetch('/personspec.json').then(r => r.json());
const { identity, voice, vocabulary, guardrails } = spec.person;

const brief = `
  Write as: ${identity.name} — ${identity.headline}
  Tone: ${voice.tone.join(', ')}
  Formality: ${voice.formality}/5
  Perspective: ${voice.perspective}
  Use these phrases: ${vocabulary.signaturePhrases.join(', ')}
  Never use: ${vocabulary.avoid.join(', ')}
  Avoid topics: ${guardrails.topicsAvoid.join(', ')}
`;

Multi-parent voice calibration

When a person represents multiple organisations, PersonSpec lets the consuming tool select the appropriate alignment per audience or platform.

// Calibrate the voice based on which parent is being represented
function getVoiceFor(spec, parentUrl) {
  const parent = spec.person.identity.parents
    .find(p => p.brandspec === parentUrl);

  return {
    baseVoice:  spec.person.voice,
    voiceDelta: parent?.alignment?.voiceDelta,
    topicScope: parent?.alignment?.topicScope,
    tensions:   parent?.alignment?.tensions ?? []
  };
}

// Use the calibrated voice when posting on behalf of that parent
const calibrated = getVoiceFor(spec, 'https://anotherco.example/brandspec.json');

Cross-platform consistency

Same PersonSpec, multiple destinations. LinkedIn drafts, newsletter editions, conference bios, podcast intros, ghostwritten articles — all produced from the same canonical voice definition, all aligned to the parent brand without manual rework.

// Generate a LinkedIn-ready post
const hook = spec.person.structure.hookStyle;      // "story"
const length = spec.person.structure.defaultLength; // "short"
const pillars = spec.person.pillars
  .filter(p => p.relationship !== 'tension')
  .map(p => p.name);

// Pass to your post generator with a chosen pillar as topic

PersonSpec is designed to be standards-aware. It builds on, references, and is interoperable with several established frameworks. It is not a replacement for any of them — it is a structured layer that makes them composable in a single, machine-readable file.