Looking at the news recently, it seems like Anthropic is Thor and Claude is its hammer. And this is the case with every AI LLM creator company.  Each day we are seeing more lightning strikes in the AI world, each day the hammer swings and another segment shatters, IT services, Equity Research, Mortgages, whatever is next. The markets are reacting to it, customers are becoming more concerned, so are vendors. What is being put at risk is trust and certainty. What is prevailing is a fear of survivability. 

‍

Has AI really achieved God status in the software realm? Or is it an illusion being cast temporarily by AI companies to pave a smoother way to that reality?

‍

These are the thoughts that are keeping teams, leaders and pioneers alike awake at night.  

‍

Before we continue, this isn’t advice for prompt hobbyists—it’s a contemplation for teams and leaders deciding how much control to give away.

‍

So what are some clear ideas we can filter out from the flood of information and speculation?

‍

Hammer vs Carpenter, who controls the future?

‍

A hammer is only as good as the carpenter who uses it. The same can be said about AI. While it does make life easier and hard work can be smashed out quickly, it is important to know how to wield it. A hammer can hurt as much as help depending on how you swing it. So much of the effort needs to be focused on evolving as a carpenter rather than just looking for the next fancier hammer.

‍

So here are some quick questions to ask:

‍

How do we pick up the right skills?

‍

Train to use the tool right. Don't focus on getting everything right, just focus on getting less wrong. It's like the stock market, novice traders focus on profit strategies, experienced ones focus on risk management for decision making. 

‍

Find ways to pick up the most common skills, prompting, RAG, Agents, etc quickly and hit the ground running. Learn along the way, but keep a focus on solely getting the simple things right and getting them to work for you. Don't worry too much about the thunder and lightning shown in the media.  

‍

Remember that AI is not automatically going to take over an entire industry, people still have to let it in and are still in control of when and how this happens. Acquire the skills and knowledge to understand these processes and decisions.

‍

How do we build mastery and not dependency?

‍

It is important to consciously decide how AI is playing a role in your future. Is AI deciding your direction—or are you still setting the vision and integrating it deliberately?

To clarify this for yourself, oversee all outcomes at all levels, at least for the foreseeable future.  Measure, monitor, validate and verify as much as you practically can. This will help you identify where you can pull back and where you can push forward with throwing AI into the mix. This is a very key decision we need to make, when things are moving so fast that it is easy to forget to check if the brakes are working.

‍

In short, stay involved in the process to ensure compliance from AI, don’t delegate and abdicate.

‍

What security and privacy means to you?

‍

Define your own privacy policy, now that privacy is no longer only limited to what is accessible to outside people but also what is accessible to inside AI models.  What are you comfortable exposing to AI and what are you not? How do you maintain the balance?  A clear idea of what these things mean to your organization can help clarify many of these decisions.

‍

What's happening in the AI world?

‍

Watch what comes next.  Simply observe and note things, don't react, only verify, explore and put down your observations.  Have an internal AI daily newsletter. Knowing what's coming up and how it fares is simply the safest way to align with it. Jumping into the water and then trying to decide how to stay afloat may not be a safe option in such turbulent tides.

‍

What's a cheaper way to get things done?

‍

Ask this question impartially, not as a cost cutting measure about how to replace people with AI and automation, but as a discovery practice to see where costs are moving and what they can eventually become. AI might be cheap or free right now, but at some point the investors will want their money back and AI companies would be hard pressed to get customers fully dependent on them and eventually have the power to hike prices against an inelastic demand.  So find ways to implement and maintain safe fall backs, a sort of hedge against AI power/capability failures.

‍

What can we build that we can own?

‍

It's one thing to make use of what's out there and make a quick jump in profits, but it is another to safeguard your future. As a Business continuity plan, you need to consider and explore possibilities to build your own proprietary models that run at a lower cost and perhaps a slightly lower efficiency but are in your control. At the very least have some strategy in place to cover some of your needs with a home grown model if not all.

‍

Where can we cut time?

‍

One of the most fundamental ideas to use for our decision making is to see where we can cut time actively. Instead of worrying about people, skills and other costly decisions, make active time compression part of your daily operational processes. Create a work culture where the team is tasked with looking for opportunities to cut time in their daily tasks and processes.

‍

Where can we replace the hammer with a stapler?

‍

There may be opportunities to replace an initially fully LLM powered logic with some specialized code. Find small openings in the workflow where you don't need complex interpretation and can use simpler methods or techniques that are more under your control. Simple code is always cheaper in terms of infra and maintenance, once it has been crystallized. You may not be able to replace AI completely, but think of AI as your fighter plane with quick response times and limited range and your overall systems as the aircraft carrier (with fuel, engineers, common centers) that can get you closer to the problems you are trying to solve and you simply fly sorties from there.

‍

Where are the simplest wins?

‍

Look for opportunities where you can use AI to solve your own problems before you use it to solve a client’s problems. For example, a support workflow or an internal ops tool that uses AI to support decision making and actual execution.

‍

Build expertise in transforming legacy systems to AI powered or AI backed solutions. It’s likely that legacy systems will not lend themselves to replacement without proper planning, oversight, validation and trust building for the users. Look at your own industry and strengths and become champions for AI in that segment.

‍

Understand where data exposure to AI is a real threat. Not all businesses may be okay sharing data and they may still be interested in using AI. Find the right balance for them. Leverage in house models and LLMs in a way where this can be achieved easily. Again, an understanding of what exposure to AI means for a particular use case may go a long way in helping move things forward. For example: do you need to anonymise identities so that conclusions cannot be drawn that affect the business or individual. Or perhaps where your needs do not require AI to interpret numbers, you can replace numbers with dummy numbers for AI to consume.

‍

Final Thoughts

‍

AI is here to stay and it is inevitable. Dependency on it is still optional, but mastery is a choice.  The right decision needs to come from a place of clarity and not fear or pressure. Asking simple questions may be a good way to get there effectively. The organizations that survive will not be the ones with the biggest models but the ones with the clearest intent.