Sidu Ponnappa
Aug 9, 2024
AI will be a step change in org velocity and productivity, no doubt. But that doesn't mean you have to start with large steps.
It actually helps to be less ambitious when starting out with AI transformation and take small, meaningful steps towards increasing velocity and throughput. Be strategic in choosing which functions to automate with AI. Focus on areas where AI can add significant value even at relatively low performance levels, rather than shooting for moonshots that work only if models perform perfectly.
Look for AI automation efforts in areas where:
The value of automation is high (relaxes other high-value bottlenecks)
The consequences of failure are low
The performance requirements are achievable with -20 points of IQ current foundational models have
For example, you can't automate outbound sales calls or zero-to-one discovery.
Sales is a high-stakes interaction that heavily relies on real-time discernment and judgment. Humans are very good at detecting when something is "off" in a real-time conversation. Plus, the reputational damage from a failed AI sales interaction would be much greater than from a failed customer support interaction.
People expect some friction in support, but not in sales.
So, be ambitious but also pragmatic. It's easy to pick the most valuable and pressing tasks and pitch them for AI automation. It's easy because these tasks stay top-of-mind. But just because they're top-of-mind doesn't mean they're good candidates for automation!
To come up with good ideas for automation, you will need to closely observe your workflows and see which tasks appear frequently and require little decision-making or judgment to execute. These are primary candidates for AI. Over time, as the technology matures, you can start to tackle higher-stakes functions.
But for now, it's wise to pick the low-hanging fruit and build trust and competence with AI in less risky areas with a standard playbook for the task and standard criteria to evaluate the output.