How we work with AI

AI as a tool, not the operator

Senior PR experts run every account. AI agents handle the research, drafting, and admin so the humans can focus on strategy, story, and relationships. Here's exactly where the line sits.

Principle 01

Senior humans own outcomes

A senior PR consultant signs off on every output before it leaves our walls. AI doesn't pitch journalists, AI doesn't approve copy, AI doesn't make commercial decisions.

Principle 02

Transparent by default

You always know which models we use, where, and on what data. Our full governance addendum lists every model, every use case, and every safeguard.

Principle 03

Your data stays yours

We never train models on your content. Every provider we work with offers EEA-located processing and zero-retention or short-retention contracts.

Where the line sits

Humans on the work that matters. AI on the work that scales.

Senior PR consultants

What our humans do

  • Strategy. Defining your narrative, picking the angles, deciding what's a story.
  • Journalist relationships. Real conversations, on the phone, over coffee, by name.
  • Final approval. Every pitch, every press release, every reactive comment is read and signed off.
  • Crisis & sensitive comms. When the stakes are high, AI assists but doesn't lead.
  • Calibration. Reading whether AI got the tone, accuracy, and nuance right - and rewriting if not.

AI agents

What our AI does

  • Research. Trawling thousands of articles, transcripts, and filings to surface the angles that matter.
  • Drafting. First drafts of pitches, releases, talking points - using your tone of voice, briefed by our team.
  • Journalist matching. Surfacing reporters statistically most likely to write your story, ranked by historical response.
  • Measurement. Pulling live data from Ahrefs, GSC, web analytics, and Brand Radar into your dashboards.
  • Admin. Brief synthesis, transcript summaries, calendar prep - the work that used to eat senior time.

Our content workflow

Seven stages, every one with a human checkpoint

Every piece of content we ship goes through this workflow. AI accelerates the middle - humans own the start, the finish, and every checkpoint in between.

Read our AI Governance Policy →
1

Strategic Planning

HUMAN-LED

Client objectives, target audience identification, messaging framework. 100% human-driven - AI is not involved.

2

Research & Preparation

AI-ASSISTED

AI conducts research synthesis and competitive analysis. Humans validate and augment.

3

Content Outline

AI-ASSISTED

AI generates structure and identifies key points. Humans review, refine and approve.

4

Draft Creation

AI-ASSISTED

AI generates initial draft. Humans conduct comprehensive editorial review, fact-checking, and accuracy verification.

5

Quality Assessment

AI-ASSISTED

AI analyses grammar, brand voice consistency, and SEO. Humans evaluate recommendations and complete final polish.

6

Client Review

HUMAN-LED

Humans present content, incorporate feedback, and obtain final client approval.

7

Publication

HUMAN-LED

Humans conduct final compliance review and handle publication. AI monitors performance.

Quality control

Four layers between AI output and what reaches a journalist

No AI-generated content goes out without passing through these. Zero tolerance for fabrication, every factual claim verified against authoritative sources.

Read our AI Governance Policy →

Layer 1

AI self-assessment

Confidence scores, flagging of uncertain content, self-citation of sources.

Layer 2

Automated quality checks

Grammar and syntax validation, plagiarism detection, brand voice consistency scoring, readability analysis.

Layer 3

Human editorial review

Professional editor review of AI-generated content, fact-checking, citation verification, tone and style alignment, cultural sensitivity review.

Layer 4

Client review and approval

All final deliverables require explicit client approval before publication.

When AI is not used

Where we keep AI out of the loop entirely

Some work is too consequential, too sensitive, or too judgement-driven for AI assistance. These are the no-go zones.

Read our AI Governance Policy →

Strategic planning and decision-making

Client relationship management

Legal or compliance advice

Final publishing decisions

Contract or business negotiations

Crisis communications requiring human expertise

Models we use

Right model for the right job

We pick the model that fits the work - not the brand. You'll always know which one we're using.

Anthropic

Claude

Long-form writing, research synthesis, transcript analysis. EEA processing via Anthropic Ireland.

OpenAI

GPT models

Short-form drafting, structured extraction, classification.

In-House

Ivy AI assistants

Specialised assistants we built for journalist matching, coverage tracking, and outreach orchestration.

Embeddings

Vector search

Semantic similarity for journalist matching, story-angle deduplication, and content classification.

Transcription

Whisper / Fireflies

Meeting transcripts and interview captures, processed only with explicit consent.

Image generation

Used sparingly

For internal mockups and concept sketches only - never for final assets that ship to media.

For the deeper read

Read the AI Transparency & Governance addendum

15 sections covering architecture, oversight, quality control, bias, explainability, data privacy, IP, compliance, incident response, and customer support.

Read the policy

Ready when you are

Senior PR with the speed of AI

Schedule a consultation with our senior team - we'll walk you through exactly how AI fits into the work we'd do for you.

Schedule a Consultation