So, what’s new about AI?

Generative AI has the potential to revolutionize nearly every industry, including healthcare, finance, and education”- Sam Altman

“Between the launch of GPT-4, which was not quite a year-and-a-half ago now, it’s 12 times cheaper to make a call to GPT-4o than the original chat or the GPT-4 model, and it’s also six times faster in terms of time to first token response.” – Kevin Scott, CTO Microsoft @ Microsoft Build 2024

We have all seen the headlines about the impact of AI on labor and productivity. Many have experimented with ChatGPT, Dall-e, Claude and other platforms. The forecast by ‘experts’ is that AI will automate our life: ask for a trip to Bali and AI will not only show you the itinerary, but will book flights based on your preferences, book the hotels, book your favorite activities (hanging tree yoga anyone?), book dinner at the restaurants providing them with your food allergies, and arrange for car transport. It is a life co-piloted by “Jarvis” from Ironman.

This agent-based action will appear across business, industry, military, personal, and other uses.

We already see glimpses of this every day. Transportation companies like Uber and airlines adjust prices based on demand. The power grid automatically adjusts flows and connections as weather, . er demand and other variables fluctuate.  Customer service provides automated responses to your questions. Etc. I.. very one of our lives these “agents” are automating tasks in hundreds of ways.

So, what’s new? Automated Learning

Automation has been a long evolution with continual improvements in logic as computers advanced in speed and power while decreasing in size and cost. Most B2B services and software companies have been leveraging automation in a myriad of ways: marketing automation, QA testing automation, customer service automation, etc. Good management teams have continually introduced these capabilities to keep their offerings and costs competitive.

Far the most part this advancement has relied on” programmatic” automation. By “programmatic” we mean based on human defined rules, based on historical data or theoretical models that automate steps in a human-defined process through “if-then ” statements. Humans use computers to implement actions designed based on conclusions either from their brains or from directed analysis on other computers. One term of art for this is ” Robotic Process Automation:

AI-based automation adds an” R” to this “R”-evolution. Automation has been constrained by individual human capacity to develop hypotheses, assess them, build software models for different scenarios, and then continually update those models and programs as new facts or inputs arose.

AI will radically reduce that human constraint by automating the logic creation itself. At their most capable, AI-agents will absorb raw, unstructured data, find patterns, develop test hypotheses, implement actions that align to the most desired outcomes, and iterate on tests to improve outcomes, all without a human involved.

Imagine a doctor seeking new patients. In an AI agent world, she would enter that request into a” chatbot” and would receive a list of new patients every week with no human in the loop.

• The chatbot would look up the doctor’s specialty.
• It would then compare known databases of patients with similar health issues or at least the demographics of those types of patients.
• It would draft marketing materials: ads, content marketing, Instagram post, TikToks, etc. using the huge public database of ads and content by similar doctors.
• It would then test different campaigns by opening Instagram and other social media accounts and posting content, placing paid search ads & other paid social posts, etc.
• It would refine the program based on most efficient campaigns.
• It would provide customers a way to sign up, even offering a live chat with a knowledgeable chatbot on that healthcare need.
• It would schedule visits for the doctor by linking with calendly or similar app.
• Then it would send an invoice to the doctor for a referral fee after the 1ˢᵗ visit – not a fee for services, advice or software licenses. But a true referral fee once the patient arrived at the visit.

This exact sequence is happening today.

This step change or “revolution” in automation offers tremendous opportunities- and threats – for B2B software+ service providers.

Potentials opportunities such as:

1. Increased efficiency in marketing and sales per the above example.
2. Increased efficiency in delivery of physical services by providing automated insights to even the most inexperienced resources.
3. Increased customer satisfaction by using AI and software to deliver an outcome rather than a service.
4. Extending the life of the most skilled technicians by turning their knowledge into software.

On average, across the three studies, generative AI tools increased business users’ throughput by 66% when performing realistic tasks. -Nielsen/Norman Group July 2023

But threats also exist:

1. Customers will come to expect outcomes rather than purchasing a software license or a block of hours.
2. Competitors will use these tools to create differentiation, and the capabilities build on themselves offering first movers and advantage in experience.

The challenges to adoption are significant, however.

• Adopting these capabilities requires a deep understanding of their inner workings and pitfalls.
• In the near term, most will require some form of “man a machine” where a human takes certain tasks, including an overall sanity check, until the agents are ready for stand-alone operation primetime.
• Applying automation requires scale. Skill aggregation, accessing models, accessing proprietary data all are improved with scale,

Therefore, there is a tremendous opportunity to scale B2B business services and software companies who have deep vertical expertise by combining that experience with advanced AI/automation capabilities. Together this “man and machine” will allow winners to reinvent their customer offerings gaining share and expanding the market through innovative “outcome” offerings currently unknown to customers.