You're Not Going to Be Replaced by AI. You're Going to Be Lapped.
A practical AI playbook for political and comms professionals
Photo by D’FramePro on Unsplash
I used to have a Post-it note on my monitor that said: “Can AI do that?”
Not because I was afraid of being replaced. Because I kept forgetting to ask. The tools existed. I just wasn’t wired yet to see my own work through that lens.
That’s where most political and communications professionals are right now. Not in danger of being replaced, but in danger of being lapped by colleagues who’ve already made the mental shift.
This isn’t a piece about robots taking jobs. It’s about efficiency — what happens when a colleague takes a two-day task and completes it in 20 minutes. It’s about what you need to learn before that AI gap becomes visible to your boss, your client, or your candidate.
Here’s what that shift looks like for three roles I watch closely.
1. The Lobbyist / Public Affairs Strategist
You built your career on relationships, institutional knowledge, and knowing which phone call to make. AI doesn’t have your Rolodex. But it’s about to make the person with less experience look a lot more prepared than they used to.
During this year’s State of the Union, I got a glimpse of what this looks like in practice. A tool called Delve AI, built by friends of mine — and one I advise — let me build custom prompts in advance for topics I knew would matter to my clients and me: tech policy, platform regulation, AI governance. As President Trump spoke, the tool transcribed the speech in real time, flagged relevant sections as they happened, and helped me generate a briefing memo on the spot.
I sent that memo around before most people had even located the official SOTU transcript.
That work used to take me hours—waiting for the transcript, pulling the relevant sections, writing it up. Now it’s a background process that runs while I’m cooking dinner or on a call.
What AI changes:
Real-time legislative and speech monitoring with pre-built client filters means you’re never waiting for the transcript again
Coalition mapping and stakeholder research are increasingly automatable — the synthesis and the relationship strategy still aren’t
Tracking regulatory and political risk across multiple issue areas simultaneously is table stakes now, not a premium service
The shift you need to make: Stop thinking of AI as a research shortcut. Start thinking of it as the analyst who preps everything so you can walk in already knowing what matters. The question isn’t whether to use AI — it’s how fast you can get to the insights it can’t generate: what this means for *your* client, in *this* political moment.
Try this now: Before the next major speech, hearing, or news event on your calendar, spend 30 minutes building custom prompts around your top client’s priority issues. If you don’t already have a real-time monitoring tool built for this — and you should, Delve is worth a serious look — here’s the bare minimum version that works today:
Bring your phone to the next event where a key figure relevant to your client is speaking
Record it using Otter.ai or any transcription app, and
Then run the transcript through this kind of prompt when it’s done.
Think in terms of specific policy topics, not broad categories. Not “tech policy” — but the actual issues your client is tracking right now. Here’s an example of what a well-built prompt looks like, using the recent wave of tech company pledges around data center electricity costs:
“Here is a transcript from [event/speech]. My client is a technology company closely tracking policy developments around data center energy costs and electricity infrastructure investment. Please do the following:
1) Identify every section where the speaker directly addresses data center costs, electricity grid investment, or energy policy commitments.
2) Flag any specific dollar figures, timelines, or company commitments mentioned.
3) Note any language that signals regulatory intent or legislative openings.
4) Summarize the top three implications for a tech company that has made public pledges around paying for electricity infrastructure. Keep the summary to one page.”
2. The Communications Director
Your job has always been part strategist, part writer, part firefighter. AI is about to make the writing part nearly invisible as a skill differentiator, which means everything else needs to sharpen fast.
But here’s what most comms professionals are missing: the opportunity isn’t just using AI to write faster. It’s building a finely tuned tool that actually sounds like your client.
The comms directors who are ahead of this are building custom AI models trained on their clients’ speeches, interviews, op-eds, and written work — then layering in a custom style guide that captures voice, cadence, and the phrases they’d never say. The result is an AI that drafts in your client’s voice, not a generic approximation of it.
Without that, you get what a lot of teams are producing right now: content that’s fast but flat. It sounds like AI wrote it because, uncalibrated, it did. Over time that erodes the thing you’re paid to protect — your client’s distinct voice and credibility.
What AI changes:
First-draft production for statements, op-eds, talking points, and social is effectively automated — but only valuable if the model is trained well
Message testing and opposition research synthesis happen faster and cheaper than traditional methods
The volume of content expected from your team goes up, because there’s no longer a labor excuse for low output
The shift you need to make: Your strategic instincts are your product now, not your ability to write under pressure. And building the AI infrastructure around your client — the custom model, the style guide, the guardrails — is now part of your job. If you’re not doing it, someone else will do it worse, and your client will notice.
Try this now: Open Claude or ChatGPT and upload 10-15 pieces of content from your most important client and make sure you’re including speaking transcripts alongside written work, because voice and written register are often very different. Then use this prompt:
“Based on these materials, create a detailed style guide for this person that captures their tone, vocabulary, sentence structure, recurring themes, and phrases they would never use.”
Save what comes back. From now on, attach that style guide to every prompt you give AI when creating content for that client. The difference in output quality will be immediate.
3. The Campaign Manager
Campaigns run on speed, resource constraints, and gut instinct. AI should be the most natural fit here — and it’s also where I see the most magical thinking happening on both ends. People either think AI is going to fully automate the task of persuading a voter base, or they think it’s irrelevant because campaigns are human. Both are wrong.
Here’s what’s actually changing: for the first time, campaigns can have a real two-way conversation with voters at scale.
Think about how much data gets lost in a typical campaign. Every door knock, every phone call, every town hall — those interactions happen and then largely disappear. Maybe someone writes a note in the voter file. Maybe not. With AI, you can now gather and analyze transcripts from every voter interaction, identify trends across your constituency in real time, and actually respond to what you’re hearing — individually, not with a mass blast from a do-not-reply address.
That last part matters more than people realize. Voters have been trained to expect nothing back from campaigns. The campaigns that use AI to close that loop — to follow up, to personalize, to actually listen at scale — are going to feel categorically different to the people they’re talking to.
What AI changes:
Voter interaction data — calls, doors, events — can be transcribed, analyzed, and used to identify real-time trends across your district
One-to-one voter communication at scale becomes possible without a massive staff
Opposition research and message testing timelines compress significantly, leveling the playing field for under-resourced campaigns
Try this now: Take stock of three things this week. First, what voter interaction data do you already have — call logs, door knock notes, event sign-ins, town hall recordings — that’s sitting unused? Second, what capture systems don’t exist yet that you need to build, even something as simple as recording and transcribing every canvasser debrief? Third, and most importantly: who on your team owns this? AI won’t process and synthesize your voter data automatically. Someone needs to be responsible for running it through regularly and turning it into actionable reports for the campaign. If no one owns it, it won’t happen.
The shift you need to make: Speed without judgment is a liability. An AI-generated voter contact script that hasn’t been stress-tested against your district, your candidate’s voice, and your opponent’s vulnerabilities can do real damage. The campaigns that struggle in 2026 won’t be the ones that ignored AI — they’ll be the ones that moved fast without building in the human layer. And given how much personal data flows through voter contact, the campaigns that think carefully about trust and transparency will have an edge with a skeptical electorate.
The Post-it still applies.
Before you start any task this week, ask yourself: “Can AI do that?”
If the answer is yes and you’re not using it, you’re spending your most valuable hours on work that doesn’t require you. And someone, somewhere, is building a competitive advantage while you do.
The goal isn’t to be replaced by AI. It’s to be irreplaceable because of how you use it.




This was so powerful!
Good stuff. Thanks Katie!