By Max Fellows, Founder of allpoints
Talk to anyone running an agency right now. They’ll probably admit that admin, process, planning and forecasting is eating them alive. You open your laptop at 8 AM. With a plan to finally nail down that company strategy, get that long-anticipated leadership post out on LinkedIn (It is going to be a good’un) or sort out your people gaps and hiring plan. Two hours later, you’re completely buried. Emails are flying around and Zoom meetings keep popping up left, right, and centre.
Then it’s all of a sudden, 1 PM. You are completely drained; you have been ‘in the business, not on the business’, yet the actual, meaningful work is sitting there untouched and likely to be an evening event. This is exactly why I am more of a believer in the benefits of AI than ever. I’m not talking about replacing our teams, faking client relationships, or losing the human experience that makes an agency great, and a culture to match. I’m talking about using AI to automate, be more effective and efficient, and lift the heavy operational burden that can pull us down daily.
BUT, as a caveat, to get it working well, the way that is most effective and efficient for your roles and requirements, it does require a time investment, the time away from your usual day-to-day in order to get it set up, trained, and deployed.
1. Stop letting operations eat your day
Microsoft recently dug into this with their Work Trend Index. Leaders are spending roughly 57% of their working hours just talking about work, sitting in meetings, writing emails, firing off messages, but the worst part was people were losing nearly nine solid hours a week just fighting their inbox. One we all know too well.
AI can actually act as an escape hatch here. Your team can use it to:
- Condense miles-long email chains into quick summaries.
- Yank action items and deadlines straight out of meeting transcripts.
- Draft up a rough, first-pass project timeline.
- Automate actions and follow-ups from calls and meetings
- Untangle messy information across tricky accounts.
- Spot missing puzzle pieces before they blow up.
- Turn tedious, repetitive tasks into quick internal processes.
Let’s look at some proof: researchers at the National Bureau of Economic Research tracked a massive group of customer support agents, over 5,000 of them. Giving those workers a basic AI assistant bumped their productivity up by an average of 14%.
Don’t get distracted by the shiny, ultra-expensive tech, either. Just figure out what boring and what repetitive job steals your team’s time every single week.
A simple but effective way to assess what your priority is (and that for the team, for example)
- What specific task took up most of your time today
- What is the key role that you do that delivers the most impact or profit for the business/team?
- What one role or set of tasks would you like to not ever have to do again if you were able?
Answer those, and you will have a simple but clear area of where to start when it comes to developing and deploying AI tools and agents.
2. Give AI a proper brief (treat it like a new hire) and be mindful that each AI tool responds and acts differently.
This will likely be a known element, but given our experience of leadership in the current market, it’s worth repeating. We’ve all heard the complaints. “AI just writes random rubbish”, “AI hallucinates” or even “The answers are so robotic”. Sometimes the software glitches out, but most of the time it’s the person typing the prompt that is giving it terrible instructions. And no matter what, you should ALWAYS check the output before sharing or acting upon it.
Imagine walking up to a brand-new creative lead and saying, “Hey, plan that event.” without explaining the budget, objectives, KPI’s, deadline and context, this is just something you would never do. AI needs that exact same hand-holding.
If you want gold out of it, your prompt has to cover:
- The Job: What exactly are you asking it to do?
- The Role: What role or experience/ skill sets are you wanting it to leverage?
- The Context: What background info does it need to know?
- The Facts: Which hard, confirmed details should it stick to?
- The Guardrails: What do you absolutely not want it to guess?
- The Output: How should the final answer look?
So, it’s time to stop saying: “Create a timeline for XYZ.”
Try this instead: “Look at the confirmed project info below and map out a first-draft workload schedule. Put confirmed deadlines in one bucket and suggested dates in another. Flag up any missing info or tight dependencies. Do not guess costs, supplier schedules, or approval times.”
AI doesn’t need you to be clever. It just needs you to be painfully clear.
3. Trust, but definitely verify
The stats around AI’s upside is where it gets interesting. A study published in Science showed people using AI banged out certain professional tasks 40% faster, with an 18% bump in overall quality. Harvard Business School found similar results with BCG consultants: 25% faster work, and a 40% leap in quality on suitable tasks.
However, there is a danger zone agencies need to be aware of. An AI will regularly give you an incredibly confident, beautifully formatted answer that is completely dead wrong. For experiential and events agencies in particular, you absolutely have to manually sanity-check:
- Budgets and commercial math.
- Venue rules, capacities, and specs.
- Supplier lead times.
- Accessibility standards.
- Licensing and legal laws.
- Health and safety protocols.
- Any client requirements.
Use the tech to get off a blank page and organise your thoughts, but never treat AI as a Senior Events and Project Director.
4. Keep the human touch (and set firm boundaries)
Just because you can automate a process doesn’t mean you always should. Agencies survive and grow on relationships. Navigating a touchy client call, giving a team member constructive feedback, or making a commercial pivot, these all demand real empathy, accountability, and lived experience. That is what leadership is all about.
Then there’s the privacy headache. You can’t just go dumping client budgets, sensitive NDAs, personal staff info, or secret creative concepts into a public AI tool without knowing where that data goes. The Information Commissioner’s Office (ICO) is super clear on this: if you use AI, you still have to play by the rules of data protection and data minimisation.
Where appropriate, move your team onto an approved business platform such as Microsoft 365 Copilot or Google Workspace with Gemini. The enterprise or paid-for space will provide you with a sandbox of security around the data you input and will not use it to train their models and inform others based on your data.
To avoid a disaster, every agency needs a clear policy on:
- Exactly which AI platforms are green-lit for use.
- What kinds of data are safe to upload (and what is strictly forbidden).
- Which outputs require mandatory fact-checking.
- Who makes the final call on a project.
- Where the line is drawn for human-only conversations.
- What are your clients’ legal parameters for use.
Start small, solve one problem at a time
The gap in our industry isn’t between the agencies using AI and the ones who aren’t, although this gap is becoming bigger all the time. It’s between the agencies using it with purpose and the ones blindly typing stuff into a chatbot hoping magic happens.
My advice with this one is to pick one annoying, recurring bottleneck. See if an AI tool can smooth it out. If it saves time and makes things clearer, roll it out wider.
AI is never going to replace the intuition you need to steer a team, reassure a panicking client, or pull off an insane live experience. What it will do is burn off the administrative fog surrounding that work, freeing you and your staff up to tackle the things that actually matter.
Keep an eye out, over the next few weeks, we’ll be sharing a lot more on the specific AI tools and real-world setups we believe can genuinely move the needle for agencies, as well as our practical AI workshops.