Writing about artificial intelligence (AI) almost seems cliche because EVERYONE is. It's reminiscent of ~2012 when "Big Data" was all the rage, but very few sources had actual use-cases. In the same way, most speak of AI and its power but without practicality.

‍

Here is the difference though ---> Big Data did not have a user interface.

‍

ChatGPT, Perplexity, Gemini, etc are the UIs (user interfaces) into AI. You needed a programmer to take advantage of Big Data. You do not need to do that for AI.

‍

The Vision vs the Reality

There are amazing things that AI can and will do. Much of this will just make software and web apps better as it runs in the background. The more tech savvy will are tapping into APIs to leverage specific use-cases.

‍

Most companies do not have the resources and know-how for this, yet. Utilizing AI today means understanding the UIs like ChatGPT. It means finding the right use-cases, designing prompts and processes, and experimenting. Most mid-sized businesses and smallers don't need to be building.

‍

They need to be experimenting.

‍

The vision could be something like using AI to scan your CRM meeting logs, allowing questions like, "How many time was our customer service mentioned on a call?  Was the sentiment positive or negative?" If you came to Everpeak to build-out this vision, our first question would be, "How is your organization using AI today?" 

Usually, our recommended step is to start with small "off-the-shelf" projects that identify real, immediate value via AI user interfaces.

‍

The AI Misnomer

It seems that much of the AI focus today for sales teams pertains to writing emails. This is misguided for several reasons.

‍

First, sales people simply don't spend THAT much time writing emails. More on that in a moment.

Second, this assumes that there is some clever hook to get a response. The data suggests that it is much less about WHAT you write and more about WHO you are writing to. You have a product or service. It doesn't matter how perfect you craft an email if it is directed at the wrong recipients.

Third, for AI to have any chance of writing an email that is more personalized or creative than a salesperson would requires access to information. Who is the contact the company, the historical context, the product we are selling, etc? This is a heavy, expensive lift. In fact, we haven't seen any company invest the vast resources required to (possibly) deliver on it.

‍

One of the reasons we believe so much time is spent talking about email writing is because it is the obvious way that many software vendors could conceivably deliver value. "We have the email recipient and some general information - we will write your email!"  But if you look at the results, they are not good. In fact, they are generally awful.

‍

So where do we see AI being most effective for sales teams today? Research/

‍

‍

AI User Interface - Beginner

Previously, we wrote a post on 5 AI use-cases for Sales Teams. The first user-case in this article was research.

‍

Now, we're going to go deeper.

‍

To keep things simple, let's use ChatGPT. Know that when we are talking about ChatGPT, we could be talking about almost any AI GPT in more general terms. ChatGPT is simply the one with the most name recognition. It's also quite good.

‍

The beginning AI use-case is finding prospective companies. The prompt matters. Using ChatGPT is similar to using Google. The individuals and companies that experimented in 2004 benefitted by having access to much better information than their competitors (and perhaps, getting on the ground floor of search advertising). There was a learning curve to Google, and we discovered that we could find better information by using tools and tricks (i.e. boolean operators).

‍

So how does this apply to Chat GPT?

‍

Let's say we are selling our product or service to food manufacturers in the Midwest. I want a list of companies. Below are three possible prompts that I could type in Chat GPT:

• Ok - Find me a list of food manufacturers in the Midwest.
  
  • It's too generic.
  • The lack of specificity means the results are minimal at seven companies, simply because so many would fit the criteria.  While they are correct, the prompt needs to be like an ideal customer profile (ICP). This means specific.
• Better - Find me a list of food manufacturers in the Midwest with 100-1000 employees and exclude contract manufacturers.
  
  • It's more aligned to the ICP of the company, and therefore, more specific.
  • By default, ChatGPT decides to give me 10 companies and they are good potential prospects.
• Best - Find me a list of at the top 30 food manufacturers in the midwest who have between 100-1000 employees. Sort them in order of employees and list the revenue if known. Do not include contract manufacturer. Provide the results in an excel spreadsheet and include the company website.
  
  • This prompt is extremely specific. The concern might be that it is too specific, but it actually provides MORE results. All 30 are viable prospects.
  • Asking for the format in excel and specifying that the website should be included is important. When prompting ChatGPT - ask for what you need! You'll be surprised how often you get it.
  • If this prompt is expanded to ask for 50 food manufacturers - guess what?  It returns 50, and the companies are good.  They all meet the criteria and would be viable prospects.

‍

AI User Interface - Intermediate

The best way to improve one's effectiveness in querying ChatGPT should not come as a surprise - practice. Query frequently and experiment with variations.

‍

You'll find different queries, data sources, and tools.  One such tools is CustomGPTs. What are they? A custom GPT is a specialized version of ChatGPT that has been configured for a particular purpose. Instead of being a general AI assistant, it’s shaped by added instructions, personality settings, and sometimes extra tools or knowledge so it can focus on specific tasks.

‍

CustomGPTs are powerful for your business and your sales team.

‍

For example:

• A sales research custom GPT could help sales reps quickly gather insights about prospects.
• A customer support custom GPT could answer FAQs and handle service tickets.
• A teaching custom GPT might be tuned to explain concepts step by step for students.

‍

Think of a CustomGPT as the same AI engine, but customized so it “behaves” and “works” in a way that’s best suited for a given job. Here is additional information from OpenAI.

At Everpeak, we have created custom GPTs that mimic the Buyer Persona for our customer. The CustomGPT replies just as the prospective buyer would. This gives sales people a real-time, real-world way to practice the pitch and objection handling.

‍

Our favorite CustomGPT is probably the Sales Research Teammate. Remember that list we generated in the Beginner section above? You can take any of these companies, enter them into the Sales Research Teammate, and get deep research that can assist in sales call or outreach. Play around with it.

‍

AI User Interface - Advanced

Being able to assemble a list of target accounts in seconds - not hours or days - is a massive increase in efficiency. However, the next step in research is to find WHO to contact at each of these companies. ChatGPT is very bad at this. It's not that they are bad - it is simply against their terms of service. ChatGPT and the other UIs do not make personally identifiable information (PII) available.

‍

So if you have the target accounts, you need to then go to ZoomInfo or whatever data service you use, manually look up the email address, and then copy it into your CRM or email. The advanced level is to automate this action by integrating ChatGPT and ZoomInfo. At a high level, here are the steps:

• Choose an Automation Platform
  
  • Various platforms can accomplish this
  • Zapier or Mulesoft is what we use.
• Select a Trigger & Action
  
  • Choose ZoomInfo as the trigger app and select an event, such as "Account Updated" or "New Contact".
  • Choose ChatGPT as the action app and select the specific action you want to perform, such as "Look up email address."
• Authenticate Accounts
  
  • Connect your ZoomInfo and ChatGPT accounts to the platform by authorizing them with your API keys or OAuth credentials.
  • If you wanted to automatically push the information into the Leads object of your CRM, you could authenticate the CRM as well and add an additional workflow.
• Configure the Workflow
  
  • Map the data you want to send from ChatGPT to ZoomInfo
  • For example, you would send company name and URL (website) from ChatGPT to ZoomInfo to generate the company list.
• Test & Activate

‍

This would allow you to query ChatGPT, push those companies into ZoomInfo, and push those decision-makers (including email address) into Salesforce or another CRM. Hopefully, this illustrates the power of the AI user interfaces coupled with some lightweight integrations can have in drastically minimizing the research required of your sales team.

‍

If you'd like to discuss how your organization can utilize AI for it sales team, contact Everpeak.

‍

‍