This conversation with Nick Handley and Chris Nightingale will likely hit a nerve. But it also landed at the perfect time.

Marketers are trying to make sense of AI’s impact on performance as every ad platform pushes harder toward “full automation.”

And Nick and Chris (co-hosts of the Prompted podcast) sit right at the forefront of that shift. They spend their days building AI tools, cleaning messy data, pressure-testing platform features, and trying to keep teams focused while ad platforms move the goalposts.

They didn't talk about AI in theory. Instead, they spoke about what it actually looks like inside a busy performance team.

And, refreshingly, there was no sugarcoating. No “AI will fix everything” narratives. But also no "we're all doomed, go back to college and train to become a plumber" doomerisms. 

So what was there? Well, if you want to know:

• What AI can genuinely automate today
• Where humans still need to stay firmly in the loop
• How to prepare your data, workflows, and teams for the next wave of changes
• And what all this means for paid search, social, and the rise of AI-driven campaigns like AI Max

Then watch the full session on-demand here, or keep reading for the written takeaways, session documents & resources, and more:

Timestamps:

0:00 - Intro
01:24 - Prompted w/ Nick & Chris
06:03 - Google's Ads Advisor tool
08:00 - Session agenda
09:39 - What can & can't be automated in PPC
13:33 - Managing AI expectations
17:18 - How to set up data, team, & processes for success
28:26 - Bot activity is dominating
30:28 - How AI is impacting IVT rates
33:18 - Real-world example AI applications
44:26 - How to build your own AI tools
47:38 - Future predictions & how to prepare
55:25 - Live audience Q&A
1:01:04 - Final thoughts

The real state of automation: what’s ready and what isn’t

A lot of automation is now genuinely useful, but usually in the boring places, not the shiny ones.

Nick called it “the heavy lifting no one sees,” like pulling reports, organising SQR exports, and stitching together data from different platforms. These are the tasks that drain time but don’t need a strategist’s brain.

But the work that clients pay for? Still human. Strategy, creative direction, and diagnosing performance swings all rely on judgment, curiosity, and context.

When Cloudflare went down recently and took half the internet with it (including GPT) Nick laughed and said it was a good reminder:

“You still need the skills in-house. AI isn’t running the show.”

Chris backed this up with a small but painful example. He tried using ChatGPT to surface insights from a client’s Excel file. It kept misreading basic numbers. Even after correcting it, the model confidently repeated the same mistake. He said it cost him four times longer than doing it manually.

It’s a good snapshot of where we are. AI can accelerate repetition, but it can’t be trusted blindly. Especially when the output sits in front of a client.

Resetting expectations inside the business

One of the bigger challenges for marketers right now is internal pressure. Some leadership teams are convinced AI can replace both agencies and headcount. Others are so wary they don’t want to use it at all. Chris sees this all the time.

“It’s either panic or overconfidence. Not much in the middle.”

The truth usually sits somewhere more practical. AI can produce quick drafts, help with ideation, or tidy up workflows. But it struggles with nuance.

For example, AI-generated creative often looks “off” in ways consumers can feel instantly, even if they can’t describe why. And when sensitive data is involved, many companies simply don’t allow it to pass through external LLMs in the first place.

Chris shared how he approaches conversations with stakeholders: he digs into the actual problem first. Sometimes the issue is process. Sometimes it’s quality. Sometimes it’s budget. But it’s rarely the case that an AI tool can shoulder everything end-to-end.

Nick made a similar point. The easiest way to push back on unrealistic expectations is to show real examples of where AI breaks. When a generated image misses the mark or a model misreads simple data, it becomes obvious why you still need experienced humans steering the work.

The rise in bot activity and what it means for paid media

With how synonymous AI and marketing is right now, it's important to remember that AI hasn't just turned marketing on its head. It's changed the shape of the web itself.

Nick and Chris touched on this briefly, but the bigger trend is hard to miss: automated content is exploding, automated browsing is rising with it, and invalid traffic is becoming a bigger part of search and social environments.

We also shared our own IVT data that paints a very real picture.

One luxury retail client saw their invalid traffic rate jump from 3.7% to around 5% when AI Max was enabled on half their search campaigns (a 35% lift in exposure). At high spend levels, that’s a painful swing.

This isn’t doom and gloom, but rather a reminder that automation influences both sides of the auction.

As AI-generated content fills search results, AI-generated clicks rise too. Marketers adopting new automated campaign types should keep an eye on what enters the funnel alongside real customers.

A quick audit goes a long way, especially if you’re leaning on AI Max, Demand Gen, or broad-match-heavy structures.

Tools built by Nick and Chris: real examples you can learn from

One of the most refreshing parts of the session was seeing what Nick and Chris have actually built. Not theoretical workflows—real tools in use today.

Script Sensei

A custom GPT created by Nils Rooijmans that guides you through building Google Ads scripts, even if you’re not a developer. Chris uses it as a teaching tool for newer team members or anyone writing their first automation.

LLM brand tracker

Chris built a system in NAN that sends brand and competitor questions to GPT and Gemini, gathers the responses, and visualises the results in Looker Studio. It’s essentially an early-warning system for how LLMs talk about your brand.

A simple use case: asking “Where’s the best place to buy a washing machine?” and seeing whether your brand as an electronics retailer shows up at all.

Impression’s internal data chatbot

Nick and the Impression team built a tool that sits on top of their BigQuery warehouse. Analysts can ask it questions like “How did ROAS change week-on-week for brand campaigns?” and get a structured, accurate pull without manually digging through datasets.

It doesn’t replace analysis—it speeds up access. As Nick put it, “It lets people spend time thinking instead of wrestling tables.”

The Non-Shite Music Recommendation Engine

A fun one. Nick built his own recommendation engine fed by data from Spotify, Apple Music, and Last.fm. It suggests one new album each day based on “vibes” and what he hasn’t listened to before. It’s a simple example of how AI, APIs, and a spreadsheet can create something personalised with very little overhead.

The Northern Agent

Nick also built an agent that checks his emails before he sends them. It keeps the tone northern and direct, but removes the accidental sharp edges. As Nick put it, “It tells me when I’m being an arse”

This is what practical AI looks like. Light, targeted tools that solve one problem well.

How to start building your own AI tools

This is where many marketers get stuck. They want to build something useful, but feel like they need to be a developer first.

Chris made it very clear that you don’t.

His approach is simple: start by understanding the problem deeply. If a workflow is eating time every single week, that’s a good place to explore automation. If the problem only happens occasionally, a tool probably won’t save you much.

Chris also stressed patience. Models make mistakes. A lot of mistakes. You’ll need to refine prompts, adjust logic, and break things a few times before the workflow behaves. That’s normal.

Nick’s advice stacks neatly on top of that. Some tasks are perfect for no-code tools like NAN or Make. Others are better handled with prompt-first platforms like AI Studio. The trick is choosing the environment that matches your own comfort level.

The bigger mindset shift: you don’t have to build “platforms.” You can build small helpers that free up meaningful time every week.

What’s coming next year

Nick and Chris both shared a sense of where paid media is heading, and none of it felt far-fetched.

Longer, richer queries

As more people use AI Mode in Google Search, queries will get longer and more descriptive. This could make intent sharper—but also less predictable.

More generative creative inside platforms

Background swaps, product scenes, headline variations… tools inside Google and Meta will keep getting easier to use. They won’t replace proper creative, but they’ll support it.

Closer attention to settings

Platforms want advertisers to pick the “easy” defaults. Nick and Chris both stressed how important it is to keep a close eye on location settings, expansion options, and bidding recommendations. Automation makes mistakes quietly.

DSA’s future remains uncertain

With AI Max growing, DSA may be on borrowed time—but not gone yet.

Across everything, one theme stood out: marketers who combine strong fundamentals with lightweight automation will move faster than those who chase full replacement.

Final takeaways

Nick and Chris made one thing very clear: AI isn’t replacing marketers. It’s reshaping where we spend our energy.

A few things to carry forward:

• Clean data is the real unlock

• Prompting is a core skill, not a novelty

• Automate the repetitive, protect the strategic

• Keep a close eye on IVT as automation expands

• Experiment with small tools instead of big platforms

The marketers who thrive next year won’t be the ones who try to automate everything. They’ll be the ones who choose automation carefully, build simple tools that create leverage, and keep thinking deeply about the work only humans can do.

Session resources & documents

Get a 14 day Lunio traffic audit
Subscribe to Prompted
Subscribe to the Paid Media Lab
Follow Nick on LinkedIn
Follow Chris on LinkedIn
Follow James on LinkedIn

Google’s new “Ads Advisor” tool
Nick’s Brighton SEO deck
Overview of prompting strategies (Google)
The rise of bot-written content (Axios)
Lunio case study w/ PLAION
Script Sensei - Custom GPT
Nils Rooijmans - AdWorld Experience Deck
Omni-Search Tool built by Chris
Duolingo AI-First failure
Klarna AI-First failure
Meta’s fully-automated vision for 2026
Kenny Skelton - AI Content Creator
Punks of AI Podcast
N8N.io
Make.com
Google AI Studio
Descript - AI Video & Podcast Editor
Gemini