Yesterday I covered the fundamentals - how to get started with AI when you're not quite sure where to begin.
But what if you're already using AI regularly and want to move beyond random productivity hacks?
During the LinkedIn Live with Helen from Squiggly Careers, we spent time on practical frameworks for people who've moved past the 'ask AI how to use AI' stage.
Here's what we covered for building AI habits that actually stick.
Projects and context folders
Once you're comfortable with basic AI conversations, the next step is getting organised about how you work with it.
Instead of starting every AI session from scratch, create ongoing projects where you can build up knowledge and refine your approach over time.
For example, if you're working on quarterly business reviews, don't just ask AI to “help me write a QBR” each time. Instead, create a project that includes:
Your company's specific KPIs and how they're measured
Previous QBR formats that worked well
The audience and their typical questions
Your industry context and competitive landscape
Build this context gradually. Each time you use AI for QBR work, you're improving a system rather than starting over.
The result?
AI suggestions that get more relevant and useful as the project develops.
I've seen marketing teams use this approach for campaign planning, HR teams for recruitment processes, and operations teams for vendor evaluations. The key is consistency - treating AI as a collaborative partner in ongoing work, not a one-off solution provider.
Expert personas
Here's where it gets interesting.
Once you've got comfortable with basic prompting, you can start asking AI to respond as specific types of experts.
Let's take product strategy. Instead of asking AI for generic product advice, try something like:
“Respond as Marty Cagan, author of Inspired. I'm dealing with competing feature requests from three different customer segments. What framework would you use to prioritise these decisions?”
Or for a marketing challenge:
“Respond as April Dunford, positioning expert. Our category is getting crowded and we're struggling to differentiate. What questions would you ask to help us reframe our positioning?”
Or as an HR business partner:
“Respond as Laszlo Bock, former Google People Operations head. I'm seeing high turnover in one specific team but exit interviews aren't revealing the real issues. What approach would you take to uncover what's actually happening?”
The AI doesn't become these people, obviously. But it does draw on their documented thinking patterns, frameworks, and approaches in a way that's often surprisingly useful.
The key is choosing experts whose thinking is relevant to your specific challenge. And if their work isn't widely known, create your own record of their work in a Word or Google doc and upload it to a new project.
The culture challenge
Now you're squeezing more value from AI. But how do you start getting value from it as a team?
Here's what people don't talk about enough: the biggest barrier to team AI adoption isn't technical - it's cultural.
Remember that agency survey I mentioned yesterday? The gap between public usage (15%) and private usage (85%) reveals something important about how organisations handle new technology.
People are experimenting with AI tools, but they're not sharing what works because they're worried about:
Being seen as lazy or even cheating at their job
Breaking undefined company policies
Getting more work dumped on them once they demonstrate they can save time
The result is that everyone's learning in isolation instead of building collective capability.
That’s why leaders and leadership teams serious about AI adoption need to both set clear policies and model the behaviour they say they want to see.
Share your own AI experiments - especially the ones that didn't work perfectly.
Discuss AI suggestions in team meetings - not as final decisions, but as starting points for discussion.
Create space for people to share what they've tried without judgment.
One team I work with starts Monday morning meetings by sharing stories on AI experiments they ran the previous week. Some work, some don't. But every experiment is celebrated because the process both builds the meta skills I mentioned yesterday.
And occasionally something new is discovered. Something that helps improve creativity, productivity, critical thinking - for everyone.
Top tip for leaders
If you're responsible for how your team adopts AI, focus on creating what urban planners call "desire paths" - the informal routes people actually take rather than the official ones you planned.
Pay attention to what people are already trying. Ask explicitly:
"What AI tools have you experimented with? What worked? What didn't?"
Then build on those experiments rather than imposing new systems.
The goal isn't controlling how people use AI - it's creating conditions where they can experiment safely and share what they learn.
Some practical approaches:
Set up monthly AI experiment shares where people demonstrate something they tried
Create shared documents where team members can add useful prompts or approaches they've discovered
Establish clear guidelines about what types of work are appropriate for AI assistance and what aren't
Invest in tools that multiple people can use rather than individual subscriptions
But grant individual licenses where there's a clear business case
You don't need the most sophisticated AI strategies to get this right. Start by making it safe to try things and learn from what happens.
Looking ahead: AI agents
Quick preview of what's coming next.
The current phase - having conversations with AI - is just the beginning.
The next evolution is AI agents: systems that can take actions on your behalf, not just provide suggestions. Think AI that can actually book your meetings based on your preferences, rather than just suggesting good times. Or AI that can update your CRM based on email conversations, rather than just drafting follow-up messages.
These tools are starting to appear now but they require the foundation we've been discussing: knowing how to give clear context, understanding what AI can and can't do, and being comfortable with iterative improvement.
The bottom line
Effective AI adoption isn't about having perfect prompts or using dozens of tools.
It's about developing systematic approaches that improve over time, creating cultural space for experimentation, and building confidence with practical use-cases.
Master this foundation and you'll be ready for whatever AI developments come next.
Wait for official training or perfect tools and you'll always be playing catch-up.
Start building these skills now. Your future self will thank you.
Ollie
These insights came from my practical session with Helen Tupper from Squiggly Careers as preparation for their AI skills sprint. If you're feeling stuck on how to get AI working in your team, drop me a line.