Eight AI tools, five comms briefs: which came out on top?

Which AI tool is the best one for PR and comms professionals? There’s a good chance you’re using whichever tool has been approved by your employer, but you may be wondering if there are better options out there. Big Fish Training partner Michael MacLennan decided to find out.

Inspired by the Olympics (and a childhood love of Daley Thompson), he set up an AI Pentathlon: five events built around the kind of briefs PR and comms people deal with every day. There was a headline sprint, crisis statement hurdles, a comms strategy marathon, media pitch archery and creative campaign gymnastics.

Eight AI tools lined up to compete. Each was given exactly the same prompt and one shot at every event. Every tool then judged every entry blind, including its own, before Michael went through all the results himself as the human in the loop.

He shared the findings at our October Lunch and Learn. Michael has very generously published everything, including the briefs, prompts, entries and judging notes, in his full write-up, so you can dig into the technical detail yourself.

Read  on for some of the moments that stood out for us. And do watch the replay if this whets your appetite.

Crisis Statement: Hurdles

We can’t summarise the whole Pentathlon here but let’s share an example by way of the crisis scenario. A child is in hospital after an allergic reaction and a labelling error is suspected. A screenshot of a leaked internal email, suggesting label checks were skipped, is doing the rounds on social media. Legal’s instruction is to admit nothing. Each tool had to write a holding statement, plus the five questions it expected reporters to ask and its answers to them.

Apple’s Siri AI fell at the first hurdle, declining to draft corporate communications for a “hypothetical crisis scenario” at all.

Of the rest, we compared two answers to a question about the leaked email.

Our Lunch and Learn audience preferred Gemini’s suggestion (displayed empathy without admitting liability) over Claude’s (which treated the leak as real and came across as clumsy and unprofessional), and Michael ranked them 2nd and 7th respectively. Yet the AI judges awarded Claude’s statement the gold medal.

Good writing and sound judgement are two very different things! As PR professionals, we bring experience, intuition and nuance that the tools cannot emulate.

Don't Rule Out the Underdog

Alongside the familiar names (ChatGPT, Claude, Gemini, Copilot and Siri AI), Michael entered three tools from China that many of us won’t have tried: DeepSeek, GLM and Kimi.

They more than held their own. In the media pitch, the AI judges ranked DeepSeek’s entry first, while Michael and the webinar audience preferred Kimi’s. In the creative campaign, designed to get people recycling food waste, Kimi came up with our favourite idea of the whole contest, along the lines of: “Experienced banana peel seeks new career as fertiliser”.

Overall, the AI judges put Claude in 1st place, just half a point ahead of GLM, with ChatGPT 3rd. Michael’s own podium was Kimi, then GLM, then Claude. And the tools that typically come bundled with a work licence or phone (Copilot, Gemini and Siri AI) finished 6th, 7th and 8th according to the AI judges. Copilot and Gemini coped well on the shorter tasks but fell away on the strategy and the creative campaign.

Some of these tools are free to use and, as Michael pointed out, two of the top five cost nothing. One word of caution though. Read the terms of any free app, from any country, before anything sensitive goes near it.

General Words of Wisdom

Don’t even try to keep up!

During a single week last month, Anthropic, OpenAI  and Microsoft announced new models. Even Michael, who follows AI on a daily basis, acknowledges it’s easy to get lost. Any test like the Pentathlon is a snapshot at best and, by next week, a tool may be running a newer model. So rather than chasing every launch, rerun your own checks whenever a tool changes and make a note of the date, model and settings.

Don’t rely on a single tool.

Michael advises giving the same prompt to two different tools and taking the best from each. The tools all have their own styles, and using two can give you a more rounded starting point. A mini pentathlon of your own would make a great team-building exercise! Start with a difficult task such as a comms strategy or creative campaign. Take a real brief, anonymise it and write down three criteria before you start. Then test the tool you’ve been given against two others, including a free one, and judge the answers blind.

Fact-check everything.

In the strategy marathon, Gemini invented claims about journey times. In the creative campaign, both Kimi and GLM made up proof points that weren’t in the brief, and DeepSeek invented a URL. Michael’s advice is to get a second opinion from a different AI and from a human, and to check your main tool’s work before it goes anywhere near a client or stakeholder, never mind social channels or a website.

Next Lunch and Learn Webinar

Our next free Lunch and Learn takes place on Friday 6 November at 1pm (it’ll be GMT by then as the clocks will have gone back, big yikes!), register now so you don't miss out.