---

If you are a product / business leader, you are constantly bombarded with AI updates — OpenAI launching Atlas, Google pushing for proactive agentic engagement, MIT report claiming 95% of enterprise AI pilots fail. As the AI landscape evolves rapidly, how do you make sense of it all? How do you separate hype from reality? How do you come up with a coherent AI strategy for your product / business? A lot of AI strategy sounds impressive, but it’s hard to actually apply or clearly understand. And sometimes, a set of tactics masquerade as strategy. Here’s a practical framework I’ve found useful, beyond the hype. This has been used at a large CPaaS company, is relevant to multiple startups, and applies to non-digital-native companies as well.

---

I think about AI strategy across five practical areas — not buzzwords, not predictions — just where AI shows up in real products and services today.

First, AI-powered experiences. This is where most teams start — AI helping users write, summarize, search, or make better decisions inside a product. Say, you are using a Customer Engagement product to send a Halloween SMS campaign, and AI helps you write copy for the message. This was fairly novel in 2023, aka an ‘excitement attribute’, but has quickly become a ‘threshold attribute’ — talk about rapid decay of delight. Nowadays, it is fairly standard for users to expect AI assistance embedded throughout the product experience. And it is not just about improving text, but picking templates, smart approvals or generating multimedia content. 

Second, AI-built systems. Both Microsoft and Google have claimed that more than a quarter of their code is written by AI. Vibe coding tools like Lovable and Vercel v0 are becoming ubiquitous. Github co-pilot, Cursor, Claude Code are routinely used by developers, enhancing their workflows. What started as code completion and code generation has shifted towards understanding and managing entire repositories. AI can help re-factor code, write test cases, do code reviews etc. Still, in most companies, the efficiency gains can’t be clearly quantified or realized — unlike big tech companies, their code stack and processes are not mature enough. This is like having a powerful car on a local road with potholes — improving the road is tedious work. But there is no choice – all your competitors are revving up, you don’t want to be left behind.

Third, AI-run operations. The practice of AI Ops, focused on IT operations , has been around for some time. This typically includes data management, monitoring, analysis, recommendations etc. Nowadays the concept of intelligent AI-run operations goes way beyond this — and includes any business operation that can be enhanced by AI. This includes operations in Support, Customer Service, HR, Sales departments. If you are a digital product company, Technology Operations will be a big part of this. The individual business function leader usually manages the specifics. Should we adopt Intercom or Zendesk for AI enhanced support? Should we introduce an AI-native conversational intelligence add-on like Gong to the Sales process? Business function leaders manage the budget, vendor selection, tool adoption related to AI for their department — just like usual. But those decisions need to be part of an overall AI strategy, to maximize impact, reduce risk.

---

These first 3 areas are about AI helping your product or team generate better outcomes. The next 2 areas are different — they are about AI interacting with your business from the outside.

Fourth, AI-discoverable products and services. If a customer asks an AI system what product or service they should use, can that AI find you, understand you, and recommend you correctly? Latest research shows that around 60% of searches don’t result in a click-through— and the popularity of AI assistants continues to explode. All AI assistants routinely recommend products and services as part of the conversation, personalized with valuable context. Just investing in Search Engine Optimization or social media campaigns is not enough. And by the way, we are just not talking popular general purpose AI assistants — if you are a software product or service, you need to show up when developers are looking for an Email API on Cursor. 

Fifth, AI-consumable services. Increasingly, AI systems are not just about finding information, but getting things done. Say, you are in Turkey, working with Claude to plan the agenda for a business trip — you expect Claude to also seamlessly book your trip from Ankara to Istanbul. In other words, Turkish Airlines needs to make its service consumable by AI. New standards are evolving — like Model Context Protocol (MCP) and Agent2Agent protocol (A2A) — to help scale this. Also, this is not just for digital products and services — for instance, ChatGPT doesn’t just recommend you a set of relevant products, it is enabling users to buy those products right in the chat experience. 

---

That’s it. Five practical areas across which you can think through your company’s AI strategy. You can decide to spend more time and effort on a few high-priority topics that’s relevant for your business, but it helps to be aware of the broader landscape. A few disclaimers below.

I am not suggesting that you need a large centralized AI org to think about, run all this. I believe AI is such a profound shift that it should not be characterized as a vertical function, but a horizontal skillset. And because of the rapid innovation in AI — it is important to develop a hype-free, practical, broad AI strategy for your company, and ensure the entire company embraces it. 

I am not suggesting that you always need to start with AI strategy before trying things. It is very important to mix higher-level strategy with specific projects, tactics. But with the rapid innovation in AI, many companies mistake those tactics to be the actual stratregy.

I am not suggesting that these 5 areas are comprehensive. As the AI landscape evolves, we need to tweak this AI strategy framework. But at this point, these 5 areas seem to provide a wide enough aperture to use, without getting caught up in buzzwords.

I am expressing some strong opinions when I say ‘AI-built systems’ or ‘AI-run operations’. Would humans still be responsible for those functions? I believe humans will increasingly oversee the work done by AI. Not writing code, but planning for it, approving it. Not creating operational playbooks, but reviewing them, iterating on them. This is a mindset shift that we need to get used to. It is not good or bad — it just is.

An important side effect of having a simple strategy framework is that everyone understands it, and has a common vernacular to talk about it. Which typically leads to better sharing of ideas and best practices. Things like AI compliance, data prep, effectiveness, ROI measurements, recruiting for AI roles — can be / should be shared across the company. You can use terms like AI Center of Excellence or AI Steering Group — but whatever you call it, don’t make it bureaucratic. It is important this “central AI effort” be light weight, nimble, action-oriented and is continously improved. I am intentionally downplaying this central function — with the hype around AI, this function could take a life of its own and slow things down with communication overhead, governance etc. While ‘responsible AI’ is non-negotiable, I believe it is vital to strike the right balance between speed and risk when AI innovation is happening at such a rapid pace. 

I’ll be sharing more on each of these areas — what they mean in practice, and how teams can apply them — so follow along if this was useful. Happy exploring — wishing you clear skies on your AI voyage 🖖.