The political consulting industry is having its overdue conversation about artificial intelligence. It is happening unevenly, and most of it is happening in the wrong order.
The wrong order looks like this: a vendor sells a campaign on a flashy generative tool, the campaign uses it to draft a fundraising email or a policy paper, the result reads like every other AI-generated email and policy paper, and someone declares either that AI has revolutionized politics or that it is a fraud. Both conclusions miss what is actually happening.
What is actually happening is that the campaigns and public affairs operations getting real value from AI are using it in three places that are almost never the places vendors lead with.
Cycle time, not headcount
Bill tracking, opposition synthesis, hearing-transcript review, regional press monitoring. Senior judgment still required. The week-long task is now an afternoon.
Signal from the unstructured
Door notes, call transcripts, town hall recordings, regulatory comment files. Converting that signal into targeting is where the real advantage lives.
The repetitive load
The seventh draft of a release at midnight. The Spanish-language proof due tomorrow. The briefing for a meeting that just got moved up. Senior attention freed for what moves outcomes.
What it does not do — and this is the part the industry needs to be more honest about — is replace political judgment. The model does not know which legislator is actually persuadable on a given vote. It does not know that the staffer who said “we’ll get back to you” meant something specific by it. It does not know which donor will be insulted by a particular ask. Those are the parts of campaigning that exist in the room, in the relationship, in the years of accumulated context.
The campaigns that use AI well will be the ones that pair it with judgment. The campaigns that use it badly will use it as a substitute for the judgment they do not yet have.
We have built our practice around the first model. We use AI augmentation across every engagement — research, monitoring, production, analysis. We do not use it to replace the work of senior strategists. We use it to give those strategists more time and better information.
That is the question every operation will have to answer over the next two cycles. Not whether to use AI. The technology is past the point where that is the interesting question. The interesting question is whether the people deploying it have the judgment to know what it should and should not do.
The campaigns that get that right will run faster, see further, and make fewer expensive mistakes. The ones that get it wrong will produce the AI slop the rest of us are already learning to recognize.