The difference between AI as a toy and AI as a teammate is not the model. It is what you connect to it.
Most comms teams have already done this. You opened ChatGPT or Claude, asked it to prep you for a client call, draft a pitch, or summarize the week’s coverage. It handed back something grammatical, generic, and a little hollow. You spent more time fixing it than it saved you, decided AI was overhyped for real PR work, and went back to doing it yourself.
I want to be honest, because that experience is real, and it is the single biggest reason PR teams stall on AI: the disappointment was earned. But the problem was not the AI. You were using it blind.
Why the blind version always falls flat
Asking a bare chatbot to “prepare me for this meeting” is like asking someone on their first morning to brief you before they have read a single client file or been given a login. Of course it is generic. It does not know your client, your messaging, your last campaign, or what actually ran in the media this week. It is guessing, politely, from the average of the internet.
Two things are missing, and they are the entire game: context, and data.
Context. The assistant needs to know your world: your messaging, your clients and their briefs, your spokespeople, your brand voice, the things you always say and the things you never say. Give it that once and every answer changes, because now it is reasoning from your reality instead of a generic average.
Data. It needs a live connection to what is actually happening, not a stale copy-paste. When your media monitoring is connected, it reasons over the coverage that ran this morning, not a guess about what might be out there. This is exactly what our own Determ MCP does, and what the morning brief runs on.
The same question, two setups
Here is the difference in one example. Same model, same question.
Blind: “Prep me for my client call.” You get a tidy, generic checklist: review recent coverage, prepare talking points, anticipate questions. True, useless, and you already knew it.
Set up, with the client’s knowledge base and this week’s live coverage connected: the same question comes back as “Two things moved this week. A national daily picked up your launch and the tone is positive, worth leading with. A competitor is gaining traction on the exact message you were about to build the call around, so I would reposition. Here are three talking points, and the two questions the client will almost certainly ask.”
That second answer is not a smarter model. It is the same assistant with your context and your data behind it. That is the whole leap, and it is how we work every day rather than a demo.
The honest part: this takes setup
None of this is free, and I am not going to pretend it is. Giving the assistant a real knowledge base and connecting it to your data is work up front. That is precisely why most teams stop at the blind version and never see the payoff: building the system spends time before it saves any. You are buying easier work later with harder work now.
That is also the line between the people who still think AI is a toy and the people for whom it quietly became the most useful member of the team. The ones who win are not the ones who found a cleverer prompt. They are the ones who did the setup.
What the real setup actually is
Forget “make an account and start chatting.” The setup that matters is two moves:
- Give it a knowledge base that knows your business, so it answers like someone on your team.
- Connect it to your data, so it reasons over what is really happening, not what it can guess.
The next post is the hands-on for the first: how to build a knowledge base for a comms team, what to put in it, and how to keep it alive. For the second, connecting your media data to the assistant, the setup guide is already up.
Start with the knowledge base. That is the step where AI stops being the thing you tried once and becomes the thing you stop wanting to work without.
If you run Determ and want your media data inside the assistant, ask us for Determ MCP beta access in the chat on determ.com. New to Determ? Start a free trial.
Ivor Bihar is COO at Determ. He co-writes this series with Claude as a drafting assistant. The product decisions, opinions, and workflows are his. Next in the series: building a knowledge base that knows your business.
