Happy Sunday and welcome to InsideAI! I'm Rob May, Founder/CTO of Dianthus. We build custom AI software for e-commerce companies to grow faster, but rather than sell it to those companies, we acquire the companies with debt. If you have an e-comm company that is looking for a good home with lots of upside, please reach out.
This week I want to discuss an idea for getting Ai projects done inside big companies. Time and time again I talk to AI people who can't get real commitment inside of their companies. It's always about POCs and experiments. Why can't they commit? Because of this curve. AI underperforms initially, and in a world of short-term thinking, it's hard to bite off something that may take two years to really bear fruit. But even if you can get buy-in from leadership, how do you get the rank and file to execute on an AI plan? A marketing or finance person deep in the org, whose job is already busy, doesn't want to take time to label data, provide feedback to models, or generally change their workflow to accommodate a two year AI project that they don't even know if it will work.
Startup sales teams have solved this problem. One of the issues in startups is that, when you start selling your first product, you usually do a crappy job. Sales comp plans are very well understood and at most good sized companies, can get very focused and complex to incentivize certain behaviors. None of that works when you are looking for your first enterprise customer. The way startups do it, then, is to comp the first sales hire or two on MBOs (Management By Objectives). If you are an executive, you've probably done it in areas other than sales, but unless you've done early stage startups, you probably haven't seen it in a sales context.
The way it works is, while you don't know exactly how to sell your new product yet, you probably do know the activities that will lead you to figure that out. So you comp your first sales rep on those activities. You might give them a bonus that is as large as the commission they would get if they hit a standard sales comp plan, but their objectives might be things like:
- Call 200 leads per week, and engage with at least 20 in conversation.
- Keep a list of sales objections and send weekly to the management team.
- Build 1 piece of sales collateral a month.
Engaging in tasks like this will help you figure out how to sell your product, and it's a big phase shift for the company when you move Sales off of MBOs and into a normal comp structure.
How does this work in an AI world? You have to change people's comp structures to have a MBO component that drives AI behavior. I can see something like:
- Contribute to building a labeled data set of 10,000 examples.
- Engage with alpha machine learning model 1 hour per week and provide feedback.
- Read one book on AI this year and give a summary presentation to the team about it.
- Write out the steps a robot would have to use to automate your least favorite part of your job.
When you compensate people on specific things, those things get done. To get AI done in most companies, you need to change the discussion and probably the compensation of the people involved.
Companies are finally starting to figure out the importance of AI, and a few have done a fantastic job of implementing it in many parts of their business. But most are still in the nascent stages. This is your chance to incent your team to be an early adopter of AI at scale. You read this newsletter because you want to know what you should do with respect to AI - and my advice is to move, and move fast.
Thanks for reading.
@robmay
Comments
Post a Comment