AI Adoption for PR Teams: Practical Guardrails, Real Use Cases, and Human-First Execution
On the "PR's Top Pros Talk…" podcast, Jon Harris, Executive Vice President and Chief Communications & Networking Officer at Conagra Brands, spoke with Doug Simon, CEO of D S Simon Media, about bringing AI into communications the right way. The headline: AI can boost creativity, precision, and speed - but only if people lead. As Jon puts it, "AI is not the pilot, you're the pilot."
Start small: focus on 1-2 clear use cases
"It's mission-critical nowadays," Jon says - but that doesn't mean you roll it out everywhere on day one. Start with a tight scope so the team can see quick wins and lose the fear of trying. That reduces the quiet anxiety many staff feel about using AI at work.
- Summarize coverage and extract reporter angles
- Draft personalized pitch outlines based on a journalist's last 5-10 articles
- Create first-draft talking points and FAQs for product or issues briefs
- Build social post variations from approved copy
Build guardrails and workflows before scale
Conagra built an internal AI function that advises marketing, legal, and HR. That structure keeps teams safe and focused. The rule of thumb: AI drafts, humans approve.
- Define what AI can draft vs. what requires human review every time
- Set approval checkpoints and keep an audit trail
- Use clean, non-sensitive inputs; never paste confidential data into public tools
- Create prompt templates that reflect your brand voice and compliance needs
Earned media: personalization at journalist speed
AI can analyze a reporter's recent work and suggest angles and references instantly. That helps you send pitches that match their beat, tone, and interest - without wasting anyone's time.
- Research: Pull a reporter's last articles and extract recurring themes
- Draft: Generate a pitch outline that cites specific pieces
- Refine: Add your news hook, proof points, and a clear ask
- Human check: Confirm accuracy, nuance, and tone before sending
As Jon notes, "You're loading the deck - and that's good - because you're not wasting time or energy." Relevance becomes a strategic advantage.
Training is non-negotiable
Access to tools isn't enough. Teams need hands-on training, real examples, and ongoing feedback to avoid sounding inauthentic. Build a cadence: weekly practice, office hours, and a shared library of prompts that actually ship work.
- Teach prompt patterns for research, writing, and editing
- Show how to fact-check and cite sources
- Practice voice alignment with side-by-side edits
If you need structured learning paths by role, see these AI courses by job.
Legal, ethics, and brand safety
Privacy, accuracy, and ownership are table stakes. Put policy in writing and make it easy to follow, not easy to ignore.
- Copyright: Avoid training or publishing with copyrighted inputs you don't own
- Privacy: Keep PII and confidential data out of consumer tools
- Accuracy: Require sourcing, fact-checking, and claim substantiation
- Bias: Add reviews for sensitive topics and audiences
- Vendors: Lock down data usage in contracts and disable training on your inputs when possible
For a helpful framework, review the NIST AI Risk Management Framework. For newsroom standards on AI usage, see AP's guidance here.
Will AI replace entry-level roles?
Doug raised the concern many have. He pushes back: the people who grew up with these tools are an advantage, not a cost center. Jon agrees - early-career communicators bring perspective, curiosity, and instincts you can't automate.
- Make junior staff AI leads for research, summaries, and first drafts
- Pair them with seniors for review to accelerate development
- Let them run small pilots and report results to the team
Your 90-day pilot plan
- Weeks 1-2: Pick two use cases, write guardrails, choose tools
- Weeks 3-6: Train the team; build prompt templates; set QA steps
- Weeks 7-10: Run pilots on live work with weekly retro
- Weeks 11-12: Measure outcomes, publish what worked, scale selectively
Metrics that matter
- Pitch acceptance rate and time-to-send
- Coverage quality (relevance, message pull-through)
- Draft time saved and revision cycles per asset
- Error rate (factual issues, tone misses, policy violations)
- Team adoption and satisfaction
Keep the voice human
AI can write a first draft. You still write the story. Guardrails keep you safe, training keeps you sharp, and a tight workflow keeps the quality high. Or as Jon says, "AI is the copilot."
From the conversation
- "It's mission-critical nowadays." - Jon Harris
- "AI is not the pilot, you're the pilot." - Jon Harris
Want more
See more interviews from the "PR's Top Pros Talk" series. Interested in taking part? Contact Doug Simon at dougs@dssimonmedia.com.
Your membership also unlocks:
