Dec 2024 : State Of Sales AI (Sold with AI - Edition 14)

Things are changing faster with Sales AI than one can decipher. This post takes stock of its current state at the end of 2024 and tries to make sense of it.

As we get ready to step into 2025, I want to look back and take stock of what has happened in Sales AI in 2024.

We started the year getting gung-ho about Generative AI. People could not have enough of it. Gartner said that it expected 60% of seller work to be performed by Generative AI by 2028. It seemed that we were soon going to be living in a very new world. Especially when it came to Sales.

Copilots were getting launched by every major company. They seemed to promise the world. And then some.

By the middle of 2024, a bit of a hush had descended on Generative AI in sales. The reports now said that those 85% of the companies that were adopting it, were actually only experimenting. Most of that experimentation seemed to be coming out of Innovation budgets. Not Sales or Marketing or other functional budgets.

However, there still was a silver lining on the cloud. And it was spelled as A-G-E-N-T-S. As in, AI Agents. The world seemed to have discovered that layers of AI could almost act like humans and do a few tasks end to end.

Like prospecting.

And just like that, the world of sales, at least the SMB and mid-market organizations, got enamored with the ‘AI SDR’.

Hundreds of AI SDR startups sprouted overnight. Like waves of Genghis Khan’s army hitting its enemies one after another. And a dozen of them even took off. Hitting revenue milestones not seen since the dot-com boom. If that wasn’t enough, Salesforce went all-in on agents.

Some of us who know the capabilities (and limitations!) of AI better than others cautioned the world to progress thoughtfully. To experiment. To adopt a ‘gradual fast’ approach when promising results shone through, but to not get ahead of themselves if they didn’t.

Some listened, and some didn’t.

As we approach the end of the year, we seem to be headed into the trough of disillusionment. Here is a sample of the results being reported right now.

AI SDRs

Jason Lemkin and Saastr recently did a poll on the success of AI SDRs.

The results? They weren’t surprising.

Most others are reporting a similar experience as well.

And Salesforce, with its big Agentforce push? It is apparently hiring “people” to do exactly what?

Sell AI Agents!

Copilots

Copilots seem to be faring just slightly better than AI SDRs. While Microsoft and others are pushing the ROI narrative, the onground reality seems to be somewhat different. HFS Research has dived deep into the topic and declared that in spite of the impressive $1B ARR that Microsoft Copilot has grown to in a year, it is not much more than a paper win for the enterprise CIOs. People are willing to experiment, but adoption continues to be a real issue.

All in all, Copilots seem to be doing better than the AI SDRs. It makes logical sense as well - AI’s big limitation right now is exceeding the 80-90% level of accuracy. Hence use cases where higher accuracy is not necessary lend themselves well to AI.

My own experience with Copilots has been less than stellar. But they seem to be useful enough.

Other Generative AI

Overall, Generative AI has everyone’s attention, but there is little impact as of yet. HFS says that while 70% of Fortune 500 enterprises claim to have adopted Generative AI, only 8% are using it at any meaningful scale yet. Others like Goldman Sachs and Gartner have now taken a more measured stance too.

In summary, we can say that some of Generative AI’s use cases have already become valuable (like AI powered research) and some more will mature over time. I think it is safe to say that Generative AI is here to stay.

Other AI

If it is not Generative AI, is it even AI? At least the world seems to be working with that mindset currently.

However, accelerated by Generative AI hype, non-Generative AI continues to make quieter progress as well. Our own work at Humantic AI spans across Generative and non-Generative AI (our core Personality prediction model is a non-Generative Neural Network) whereas some of our other models used to create content are Generative in nature.

Call analysis (not all of the AI here is Generative), forecasting etc. are other areas where non-Generative AI continues to be useful.

Summary

We are in an exciting but early phase of AI. Expecting too much too soon, or going for quick wins would be perilous. However, staying put, experimenting too little and not taking some educated bets would be even worse.

Sales leaders should consider prioritizing AI that is either semi-autonomous or assistive in nature. They should experiment with autonomous AI at a smaller scale; or otherwise pick up use cases that do not demand perfection (research, coaching and so on).

The future is bright, but the sun will take its time before fully shining through.