Artificial intelligence is continuing to grow and venture into new territory. According to stateof.ai the developments in Natural Language Processing surpass anything we’ve seen before.
With prominent examples, such as for instance GPT-3, the possibilities for unlocking new business value seem nearly endless. This specific system can be used to produce software from human language, (“Create a website, with a banner and a red button”), write blog posts (not this one) and complete rows in an Excel sheet by understanding the first few human entries.
Specifically, the rise of new and more powerful Natural Language Processing (NLP) hugely impacts marketing and sales.
Sometimes it is hard to distinguish fact from fiction with buzzwords like “artificial intelligence”, “machine learning” or “algorithms” flying around. So, is there real business value in AI outside the Giants like Google, Tesla, Amazon, Apple and Facebook?
What can AI do
Looking at where AI can deliver value for Business, it falls into either of the following three categories:
- Automating business processes: For example automatically completing customer records or reading legal documents and extract the critical provisions.
- Gaining insight through data analysis: Here AI is used today to dig through vast amounts of data to detect patterns and even interpret it’s meaning. Think of analytics on steroids that can help identify a customers’ likelihood to buy, detect fraud or gather information about what type of ads to show.
- Engaging with customers and employees: Think chatbots and recommendation systems. This category represents the smallest of the three.
What AI can’t do (today 🙂
Although things are evolving at an incredible pace, AI is still very limited and a general AI is still far away. Although we have the tendency to anthropomorphize just about anything, including recently added to the list Viruses, AIs remain programs that run on computers.
Limitations of even the likes of GPT-3 become visible when asked about historic events that could not have taken place.
Q: Who was president of the United States in 1700?
A: William Penn was president of the United States in 1700.
Q: Who was president of the United States in 1650?
A: Richard Bennett was president of the United States in 1650.
Q: Who was president of the United States in 1620?
A: James I was president of the United States in 1620.
Q: Who was president of the United States in 1600?
A: Queen Elizabeth I was president of the United States in 1600.
Reasons and examples why AI is great for HubSpot data
To get immediate value out of AI, it has helped me to consider the following:
- AIs are helpers that can conduct specific tasks faster and better than Humans
- There are many tasks that are important yet not urgent (Eisenhower Matrix), therefore never get done
This means you could save time and ensure long-term compounding actions are taken automatically.
What are the things you could and should do regularly that would help you in the long run?
Here are a few that come to mind:
- Automate questions and requests processing.
- Automate customer records for user segmentation.
- Analyze customer requests.
- Prioritize customer requests.
- Anticipate trends and emerging customer needs.
- Extract customer satisfaction (instead of or complementing NPS surveys).
- Generate FAQs.
- Proactively send customers relevant information.
How to make it work
So, does it make sense for a small business to automate, analyze or enhance support using AI? This one is hard to answer, but my general answer is that even for a small business, the ways in which AI can be put to use also for non-tech organizations are increasing. The reason being the productization of AI, meaning you’ll never have touch a line of code to make use of AI.
1. Use it in your spreadsheets
Tools like text2data provide AI sentiment and categorization right from Google sheets.
Pros: Easy to use, free trials available.
Cons: Data needs to be exported and formatted first.
2. Install a HubSpot App
Pros: Easy to use, everything on autopilot, reporting, free-trial.
Cons: Makes everything accessible to the entire team.
3. Build your own
A large number of APIs are available to develop custom solutions. Google for instance provides a natural language api that lets you detect sentiment, entities, and language.
Pros: The most flexible solution to account for every case you can think of (within the limits of the API of course)
Cons: The technical know how to build an application is required. Cost and time of implementation.
AI is getting increasingly more powerful and accessible. Education and experimentation is abundantly available. The upsides of getting started far outweigh the possible downsides.
You don’t need to code to put AI to work for you!
Stay in the know
Get the tools, content and resources for Customer-centric teams straight to your inbox.