Caio and Marina take apart the idea that buying an AI tool fixes a disorganized company. The truth? AI on top of chaos just makes the chaos happen faster. Why a bad process plus AI becomes a bad problem at scale, what McKinsey found about redesigning workflows, and what to do before you sign one more AI license.
In this episode
01 Hook: the new tool that made everything worse
- Caio opens with a real scene: a company buys ChatGPT Enterprise for everyone, six months later nobody can say what got better
- Marina asks what listeners are thinking: 'but wait, wasn't AI supposed to solve exactly this?'
- Central thesis, no sugarcoating: AI doesn't fix a bad process, it just makes the bad process happen faster
- A grabber to keep you hooked: if your workflow already leaks money, AI just leaks it faster
02 Why this happens in real life
- The difference between buying a tool and changing how you work — most people stop at the buying
- Concrete example: support that already answers wrong now answers wrong in seconds, at volume
- Marina pulls the culture thread: 'so it's not a tech problem, it's a management problem?' and Caio agrees but adds a twist
- The mistake of measuring 'adoption' (how many use it) instead of results (what actually moved in the numbers)
03 The data that changes the conversation
- McKinsey: redesigning workflows is a key factor for capturing value with AI — it's not a detail
- High-performing companies are nearly three times more likely to redesign individual workflows
- Marina translates: the people making money with AI changed the process, not just the software
- Caio explains why: AI exposes the bottleneck that was always there, only now you can see it
04 How to spot where the mess is
- A practical question to ask before buying anything: does this process work without AI today?
- Red flag: if nobody can sketch the workflow on paper, AI won't guess it for you
- Sales example: outdated CRM, now with AI generating pretty reports on top of dirty data
- Marina raises the honest objection: 'but redesigning a process is hard and slow — can you start small?'
05 The turn: the path that actually works
- Caio flips the order: first fix one workflow, then add AI to scale what already works
- Start with just one process, with a clear beginning, middle and end, and measure before and after
- The leader's job: define what counts as a result, not just hand out tool access
- Marina sums up the insight: AI is an amplifier — it amplifies what's already there, good or bad
06 Practical wrap-up
- Three questions to bring to Monday's meeting: is this process clear? Does it work without AI? How will I measure it?
- Caio gives the blunt advice: pause your next license purchase until you can answer those three
- Marina closes with the line that sticks: don't scale the mess — organize first, scale later
- An invite to the next episode and to reach out to Caio for anyone who wants to actually apply this