By Matthew Dybwad, Chief Revenue Officer, IQM

Before this election is over, somebody is going to tell political buyers to finish strong.

I suspect they’ve thought of that.

The more useful question is why finishing strong still requires so much work that has very little to do with political judgment. Translating a strategy into platform settings. Rebuilding audiences. Chasing down why a campaign won’t scale. Reconciling reports before anyone can decide what to change.

Political is hard. Moving a sound decision from a buyer’s head into a live campaign shouldn’t be this hard.

By 2028, I expect political buyers to plan, execute, measure, and optimize campaigns from inside ChatGPT, Claude, or whatever agent they choose to work with. The technology underneath will matter enormously. Buyers will have far less reason to spend their days inside its interface.

BYO agent. Use your own judgment.

Imagine a buyer asking their agent how the campaign is performing against its objective. The agent pulls current campaign data, identifies the constraint, and lays out a proposed change with the tradeoffs. The buyer can see what would change, what it would cost, and what requires approval. Once authorized, the change goes to the execution system. The results come back into the same conversation.

The context survives. So does the audit trail.

Buyers also retain control over budgets, permissions, and the decisions they aren’t willing to delegate.

The engine still has to run

A general-purpose model can help interpret an objective and call a tool. It doesn’t generate a live identity graph, licensed data rights, inventory access, or a functioning real-time bidding operation by being asked nicely. Somebody still has to bid, target, pace, manage frequency, and return trustworthy performance data. An eloquent answer built on stale or poorly matched data is still a bad answer. Eating more tokens won’t make it better.

Compliance has to travel with the campaign

A buyer working across states shouldn’t get surprised by a new compliance requirement only after a campaign is ready to launch. Nor should switching from a DSP screen to an agent strip away the checks and verification that would have applied in the UI.

DSPs have a choice to make

For a DSP, this future demands a choice about where to put engineering effort: protecting the interface, or making the underlying capabilities available wherever buyers want to work. Political buyers will still need judgment, creativity, and the ability to make a hard call with imperfect information. Those are good uses of talented people.

Moving the same information through five systems is not.

We intend to move

At IQM, our direction is clear. We want to earn a place in the buyer’s workflow through the quality of our execution, our data, and our political expertise. We don’t need to own every screen they look at.

We’ve already opened a path through MCP integrations documented for ChatGPT and Claude. Underneath sits the same bidder, tracking, pacing, and targeting, connected to our data spine and identity graph. We’re building compliance into execution, with maintained rules, jurisdiction-aware checks, approval gates, and records of who authorized what. And we’d rather hand customers a tested prompt than another screen and a webinar.

Connecting an agent is a starting point. Making it useful through a full political cycle is the real test.

Finish this election. Then let’s build a buying experience that doesn’t make the next one harder than it needs to be. Fill out the form below, and we’ll show you where we’re headed.