On Speech Engineering

To paraphrase the old line about people and dogs: the more I study algorithms, the more I love people.

Working with AI models, I learned something: to get a good result, it isn’t enough to just “ask a question.” A whole new domain of knowledge has appeared — prompt engineering. It’s a set of methods: how to frame the task, gather the data, set the context and the format of the answer, so that you get specifics and unique value instead of hallucinations and boilerplate. There’s even a proverb for it now: “shit in — shit out.”

After a while I started trying to carry these principles into live conversation. And the results turned out to be surprisingly strong — especially in business.

In short, the logic goes like this:

— set or confirm the role (“imagine you’re the best analyst / lawyer / chef…”);
— lay out the context (the facts, the history, why it matters);
— spell out the expectations (format, metrics, style, deadline);
— give examples to make it concrete;
— name the constraints, if there are any;
— clarify the way of working: checklists, stages, interim results, clarifying questions.

And if the task is big — break it into subtasks and go step by step.

The art is to make the request as detailed as possible while keeping it short. The more excess information, the more your counterpart’s focus blurs, and the higher the risk of “result degradation.”

AI neural networks are built in the image of our own. Often what works “over there” works perfectly well “over here” too.

You might object: “Won’t this make speech too artificial? Won’t something living get lost?” No. Because what decides things is emotion, tone, and body language. Structure doesn’t kill sincerity — it amplifies it.

I’ve noticed that many politicians and leaders work exactly this way. First the historical background, then the current situation, then comparisons and examples, then expectations and wishes, and finally questions that leave you something to chew on.

Building a dialogue like this isn’t easy. You have to go through the inner work yourself first, to give your thoughts a structure. At times, especially at the start, it takes a fair amount of time. But soon it becomes a skill. And the result is clearly worth the effort: your thoughts get sharper, your speech clearer, and your image in other people’s eyes carries more weight. Far less is left unsaid, far less chaos, far less disappointment.

Speech engineering isn’t about “dry formulas.” It’s about understanding each other better and building a new quality into our business and personal lives.

Here’s to all of us becoming masters of communication — getting the results we want while keeping the warmth in our conversations! 😎