Ideas for AI in Sales Software

Products I'd pay for

These past two weeks I’ve been bombarded by sales pitches for sales software from people who’ve never done software sales. While there’s nothing wrong with bringing a fresh perspective to a space, it’s giving me a lot of heartburn watching smart AI founders try to tackle stale problems like automated outbounding. 

In an effort to save these folks some time, and to selfishly find solutions I can buy, I’m starting this free ideas series. I hope it’s helpful for founders who are in ideation mode and for buyers who have ideas to share. There are tons of opportunities for go-to-market tooling to be reinvented by generative AI, and I’m excited to use the products that emerge. 

Existing landscape

There are literally thousands of sales tools out there today. I dug up the market map below, which includes not just SaaS, but the tangential services and outsourced staffing that sales teams pay for. 

Billions of dollars are spent on at least 30 different categories of tooling, and you could very well pick a category and try to build an AI-powered version of something that exists. Sure, an incumbent could stand up a team and build in AI natively, but there’s a reasonable chance you could move fast enough to displace them. A lot of these tools weren’t built by product-forward teams; it’s sales SaaS after all. 

Sales process today

A better way to approach ideation is to dive deep into the sales process. It’s mostly human-driven today, and highly variable depending on the size of the company, the customers they sell to, etc. The flowchart I made below is based on my view on the process, so it won’t necessarily apply to the group you decide to sell to. Make it an exercise for yourself to create your version of the flow *after* you have a few dozen customer interviews or case studies under your belt.

Reinvent

Now that you’ve got both the eagle eye perspective (via market map) and an on-the-ground view (via sales process), you can combine the two to see which parts of the sales process have ample tooling right now. From here, there are two approaches: You could choose to go where the tools form clusters, or you could try to tackle greenfield areas. Let’s first look at the existing tools and how AI might disrupt them. I’ll focus only on the highlighted areas.

1. Prospecting tools - The ultimate infinite feedback loop 

Find the “global minimum” of messaging, ie what types of messages work for the different personas you’re hitting up? What does the perfect sequence look like, step by step? The ideal tool doesn’t even need to consider lead generation. As long as the BDR can input a spreadsheet of contacts that fit within a certain persona, the AI tool should be able to segment out that list, test out different combinations of sequences, and iterate every day, every week, or at whatever cadence works. 

It’s way harder than it sounds, because sequences often touch different channels (LinkedIn, Twitter, email, call, etc) and can be anywhere from 1 to 10 steps. What is important is that AI automates this process of finding the true message fit. Every cycle it should be looking at past cycles’ response and meeting booking rates, and using that to inform the formation of future sequences. It’s a continual process and there won’t be a definitive answer to “what is the perfect sequence,” but at least this tool would save 10-50% of a BDR’s time, which can be diverted to live qualification calls and smarter activities. 

2. ABM - Campaigns on autopilot, also on infinite feedback loop

Similar to the above, it’s easy to generate a target list of companies and contacts, but hard to test out and align messaging for those big accounts. There’s huge opportunity to make the process of ABM more scientific, from the sequence of personas to touch (ie is it better to reach out to the product manager first, or their technical counterpart?) to the channels to prioritize (should Big Buyer see your Google Search ad before they get an email?)

3. Data enrichment - Fresh data that triggers actions

An ideal AI qualification tool should be deeply integrated with data enrichment APIs and real world news and context data. Being able to detect these changes and run automated qualification and handoff to BDRs would be key. This is assuming the lead was not qualified on a call and/or qualification needs to happen via email.  

4. Marketing automation - Smarter lead nurturing

Similarly, AI-powered lead nurturing would be much smarter than current BDR response templates. Being able to tap into the sales team’s collateral and the information that’s available on the company’s website would help train AI on what information to surface when progressing through nurture phases. 

5. Call analysis - Live, on the fly direction 

Existing call recording tools are really only good for retroactive analysis and coaching. There’s space for AI-powered live call assistants for sales teams. It should be able to provide real-time insights and recommendations (perhaps via Slack pings) during product demos and customer interactions. 

6. Sales enablement - Universal search bar

AI can help sales teams quickly access relevant information, resources and training through better search across tools. Currently the bottleneck is in direct integrations between tools, ie outbounding tools sometimes don’t surface all interactions into the CRM, where a rep might be looking prior to a call. There are interesting apps like Scratchpad that are essentially like Superhuman for Salesforce, but the ideal tool is a pure search bar for the sales stack. Something like Commandbar or Slite’s (YC W18) new Ask product could work.

7. Revenue operations - Make data queries in plain English

All great sales leaders have a great revenue operations counterpart. Having a tool that can better automate forecasting and sales analytics would drive better results for sales teams. This might look like the AI assisted SQL query tools that are cropping up.

Invent

Greenfield opportunities exist because there are areas that can’t be productized without AI. 

1. Post-human call execution - Why do we even need humans on sales calls?

Not to sound too Black Mirror, but there’s a world where software buyers no longer need to interact with sales reps. Maybe you feed AI the headshots of your existing sales team or it generates its own sales reps, or in the most dystopian version, sales calls are just run by a pulsing white orb… There are so many ways we can cut down the time we spend on qualification and discovery Zooms. Obviously though, this type of tool isn’t for everyone, certainly not for the pure PLG (product led growth) crowd. And more importantly, for this to exist, we need to assume a world where building human rapport no longer matters / produces an edge.

2. Demo generation - Automated sandbox provisioning

Demos are where sales engineers spend a majority of their time, asides from developing SOWs and triaging bugs with engineering. For most products, there aren’t many combinations of factors to consider for setting up customized demos and sandbox environments. AI could reasonably automate both processes with enough context on the prospect. It could even generate or guide demo calls, similar to above.

3. Collateral generation - Generate anything a prospect will ever need

Today, product marketers grind away to address the needs of sales teams. Tons of questions arise throughout the sales process, everything from competitor feature comparison to build vs buy considerations. All of this requires combinations of PDF, landing page, slideshow, and video collateral that’s created, remixed, and sent over to prospects. AI powered collateral generation would be awesome. If trained on existing collateral (including landing pages), it could completely remove the menial work that creating collateral involves today.

4. AI concierge - No more 50+ email threads  

An AI concierge that is added to every email chain post-qualification and pre-close could be huge, saving sales reps hours every week in answering common questions and addressing sales process needs. The concierge could be trained on the wealth of data in both the prospecting tool (Outreach) and the CRM (Hubspot).  

Traps

At the end of the day, you’re building sales software for salespeople. You’re helping them focus and reach their target audience, but don’t forget that you need to do the same too! So, a few key considerations before you start grinding on your sales SaaS startup… 

First, make sure you have your ICP (Ideal Customer Profile) nailed down. I might not be in your ICP given the software stack I use and what channels my organization sells through. Every customer has a unique:

• Sales channel - Some teams lean heavy on cold calls or Twitter DM’s, and some don’t. It all depends on where their prospects spend most of their time online. The prospecting journey can really diverge depending on the initial channels messages are sent through.

• Team - Headcount, composition (ie sales engineer to AE ratio), layering, etc can all contribute to differing software needs.

• Perspective on tooling - I view my core stack as CRM, Prospecting, and Call Analysis. Some organizations might have a more complex view of what’s “core” to their work, or have a different conception of the stack completely. My specific tools are Hubspot, Outreach, and Avoma. There are other common combinations, like Salesforce, Salesloft, and Gong. 

That naturally leads to my next point - integrations. It’s unlikely you’ll build a new source of truth for sales teams, so you need to consider what parts of the stack you’ll integrate with first. Stack rank and make sure you know what table stakes are. I’ve had a few pitches in my inbox I liked, but couldn’t use because they didn’t have Outreach.

Finally, if you’re trying to build a venture-backed company, the question will always be - how do you build a billion dollar business from this? How are you going to eat into the rest of the sales stack? I’ll leave this as an exercise for the reader.

If you're a builder, comment if you’re building something! Drop a link or let me know if I’m missing the type of tool you’re building.

If you're a buyer, please poke holes in my framework and help fill in the gaps. If you’re also thinking about buying software, let us know what your key considerations are.