Civic AI - Social Media Drafting

Fabric Score 4.0
PublicPublic ServicesCivicGenAI modelExecutive / Org Leadership ×1

Workflow Diagram

Civic AI - Social Media Drafting workflow diagram

Fabric Score

ValueEfficacySecurityRiskExternalitiesEfficiency4.54.34.72.74.03.84.0Fabric Score

Task

Assist with drafting social media posts that align with a legislator’s voice, intended topic, and required word or character count using approved contextual sources.

Intent

Legislative offices publish frequent public-facing updates across social platforms. The AI system supports staff by synthesising approved context from prior posts, uploaded news stories, and the office’s existing data room to produce draft social media content that is on-message, appropriately scoped, and consistent with established tone, while maintaining human control over publication.

AI Workflow

Input

Context documents including prior social media posts, uploaded news articles, press materials, and the legislator’s existing data room, along with a possible prompt from the staff member.

Process

Relevant context is selected and analysed to understand prior positioning, tone, and factual grounding. The model generates a draft social media post tailored to the specified topic and length constraints, ensuring consistency with historical messaging and approved materials. No content is published automatically.

Output

A draft social media post that meets the specified word or character count and reflects the legislator’s established voice, prepared for staff review and modification.

Human Oversight Level

Human-Led with AI Assistance

Institutional Oversight Examples

  • Models must be fully closed, secured, and incapable of leaking data
  • Drafted content is assistive only and subject to staff review prior to publication
  • Use is governed by internal communications policies, approval workflows, and applicable public records and data governance requirements

Risk

Inaccurate or outdated statements, misalignment with office messaging strategy, inappropriate tone, or unintended public interpretation.

Output Modification Telemetry

  • Unfaithful outputs (policy, tone, or style misalignment)85%
  • Missing or incomplete information5%
  • Irrelevant information5%
  • Internal inconsistency5%

Transversal Metrics

Grouped by Fabric dimension.

Efficacy

Accuracy85 %

Efficiency

Modification Rate20 %
Modification Time1 min/output
Verification Time1 min
Rejection Rate5 %
Operational Friction5 hrs/wk
Implementation Overhead100 hrs
Governance Overhead40 hrs
Time to Launch8 wks

Value

Effort Reduction95 %

Risk

Reliability5 incident/month
Autonomy100 %