I’m finding the current state of AI sales-tech solutions to be both exciting and terrifying - well, once you dig past the ChaGpt-wrapper bonanza that is.
As a senior sales operator in a relatively large org, I believe internal sales automation workflows are something enterprise companies should prioritize.

I mean, it just makes sense if you have the data to train a custom LLM. Implementing generative models trained exclusively on publicly trained data is a major limiting factor in sales optimization - Chatgpt is still limited (it all sounds the same) regardless of how many hours one spends with prompt engineering.

That being said, what are your go-to AI tools and workflows? Curious to know what everyone is working with. Feel free to share use cases!

These are a couple of battle-tested AI sales workflows we’ve had success with so far:

Multimodaldev

We wanted to create internal generative AI and CRM optimization workflows trained on our own data lakes (copious amounts of data points on sales activity and comms).
Closed-source LLMs (ChatGpt and a couple of GPT-wrappers) didn’t cut it and it actually damaged our sales numbers. We ended up having two custom AI bots built via Multimodal:

  1. A sales comms generative AI agent segmented for each stage of the sales cycle with custom-built personas. The customization has massively improved cadence and general stylometry (i.e. the generated AI text doesn’t sound like an Englishman from medieval times or an undercover mall cop anymore). From email replies to ad copy this has been a major improvement.
  2. We have also built an atypical lead scoring/predictive analytics bot trained on email metadata and sales cycle outcomes (ie. Correlation between the # of email opens, the time interval between them, additional metadata, and sales cycle outcomes (won, lost, etc). This is very much a work in progress with reinforced learning.

Humantic
The factoring of personality scores as an indicator of success in B2B sales is still very niche. It takes persistent fine-tuning at the onset and relentless multi-variate testing across a broad audience of leads but this vector has helped increase LinkedIn conversion rates.

The only two constraints that pain me are:

  1. It only seems to work well on Linkedin - not channel agnostic from the get-go.
  2. You need to factor in Linkedin engagement volume/frequency as a filter for further segmentation - it only seems to work well when targeting LI “influencers”.