Apptivate
Get in touch

Insights

What does it mean to be AI-native?

Apptivate6 min read

There's a lot of talk these days about being "AI-native". There's also talk of being "AI-pilled". The AI industry is full of influencers and companies claiming that they have all the answers. The truth is, they don't, and we don't either. All we can do is stay on the bleeding edge of advancements, stay curious, and not be afraid of challenging the status quo.

So what does it mean to be AI-native? And can you use the term to describe a person? A team? A process? Something is AI-native when it was designed from the beginning around artificial intelligence, rather than having AI added later as an extra feature. A customer service process that is built from the ground up with swarms of AI agents is an AI-native system.

It can also be used to describe a person or group. A person is AI-native when they always think AI-first. They reject the traditional way of doing things, and try to do it with AI-first. For example, instead of calling multiple specialists after getting a referral, they offload the task of setting up an appointment to their agent.

Designing AI-native systems

Machines can now read almost anything: a scanned form, an email thread, a photo of a whiteboard. They can produce a decent first draft of almost anything: code, a summary, a filled-in record. The key word is 'decent'. AI is not magic and it will not immediately solve all of your problems.

Within the limits that you define, they can act: call a system, send a message, open a ticket. When designing an AI-native system, the list of possibilities is nearly endless and growing every day. But the question of human-computer interaction is, and will remain, a sticking point for many years to come.

How do we design reliable systems that offload most of the work to AI, while keeping the intent, priorities, high-level judgement, and in some cases, final approval in the hands of humans?

Anthropic describes its own engineering this way: Claude authors about 80% of the code merged into its codebase, while its human engineers focus on "directing, setting intent, and owning final approval".1 In 2025, Apptivate's founder experienced this new way of working when building apps and other AI systems for his friends and family.

The grunt work of coding has essentially been "solved", but planning and prioritization, taste and judgement fit firmly in the hands of the human controllers.

Gartner expected most enterprise applications to have an embedded assistant by the end of 2025, and describes those assistants as ones that "depend on human input and do not operate independently".2 The problem is, an AI assistant slapped on top of an unchanged process is just a slightly more efficient way to do the old work.

But the old work was the problem. Building AI-native systems forces you to reject your old processes and rethink the work that you do from the ground up.

AI-washed, AI-enabled, AI-native

There are at least 3 levels of applying operational AI in your organization. AI-washed, AI-enabled, and AI-native.

AI-washed is essentially an old product with some AI features slapped on top. Gartner calls it agent washing: "the rebranding of existing products, such as AI assistants, robotic process automation (RPA) and chatbots, without substantial agentic capabilities".3 There are plenty of examples of this in practice. As an exercise, open a random app on your phone and look for some AI features that were added within the last few years. Most are pointless AI chatbots that were built more for pleasing investors than for improving customer experiences.

AI-enabled is real AI added to a system or process that predates it. Take, for example, the Gemini AI features added to Google Docs, Sheets, Drive, etc., over the past few years.4 The core system is unchanged, the process of creating documents is unchanged. But if the generative AI features like text editing and document summaries vanished overnight, the overall service disruptions would be minor.

AI-native is designing a new feature, product, or system from the ground up with AI at its core. Anthropic's Claude Code is a great example of an AI-native product that enables a brand new way to code. It was built from the ground up with AI at its core, and if you take away the AI, it ceases to exist. AI-native products like Claude Code also change how teams work. The AI-native SDLC is turning out to be a very different way to code, allowing companies to build at a pace unheard of in the past.5

As an exercise, let's look at a business problem and architect a solution in 3 ways.

A pharmaceutical company creates sales and marketing materials for its various brands. Its content supply chain starts with a global team creating a campaign brief, getting it approved, then sending it off for asset production. After assets are produced, they need to be approved by global medical, legal, and regulatory teams for final approval before they can be eligible for localization to markets. After localization, the assets need to be approved again by local medical, legal and regulatory before finally being eligible for distribution.

This process includes many complex systems and spans across multiple teams and geographies. Let's suggest a solution in 3 different ways.

  1. AI-washed: give stakeholders a chatbot that they can use to ask questions about the materials and process, and give some high-level guidance to reviewers. Overall time to market is unchanged, but management feels good about their AI token usage going up.
  2. AI-enabled: Each step of the content supply chain gets a tool that helps reduce time spent incrementally. The overall process is the same, but the document text is now translated with AI instead of human translators. Overall time to market is cut by 5 to 10 percent.
  3. AI-native: The entire process is reimagined. Critical questions are asked like "Why do we have 2 separate rounds of review? Why do we rely so much on external agencies? Can we create localized versions from the beginning instead of relying on the global-to-local process?" Based on these challenged assumptions, a new process is introduced which cuts time to market by half.

How to be AI-native

Let's be honest, most teams and processes do not become AI-native overnight. The bigger the company, the longer it usually takes to change. Becoming AI-native is a cultural problem as much as it is a technology problem. Meaning that whatever route you decide to take on your journey to becoming AI-native, it must include investments in encouraging AI adoption within your company.

The most AI-native companies hire curious, independent, and persistent people. These are deep thinkers who are constantly looking for ways to improve themselves and their work. If you are lucky enough to have people like this on your team, use them as change agents. Give them freedom to experiment, and give them tools to experiment with.

For large organizations, becoming AI-native usually involves recruiting external support. Large, traditional companies often lack the internal technical support to allow the business to build meaningfully impactful systems. Many companies are not historically software product-oriented, meaning that collaboration between engineering and business does not make room for much innovation.

The business often has a plethora of great ideas of how to improve processes, but they need a partner to help them take it one step deeper. Instead of improving processes, they should be reimagining them. Then comes the question. Who will build the thing?

Apptivate specializes in reimagining systems based on core business problems. Instead of improving processes, we prefer to discuss how things could be done differently. Faster, more efficient, higher quality, fewer defects. When it comes time to build, we are there to make it real.

References

  1. Anthropic, "How Anthropic secures its AI-native software development lifecycle", blog post by Jason Clinton, 21 July 2026. https://claude.com/blog/how-anthropic-secures-its-ai-native-software-development-lifecycle ↑

  2. Gartner, "Gartner Predicts 40% of Enterprise Apps Will Feature Task-Specific AI Agents by 2026, Up from Less Than 5% in 2025", press release, 26 August 2025. https://www.gartner.com/en/newsroom/press-releases/2025-08-26-gartner-predicts-40-percent-of-enterprise-apps-will-feature-task-specific-ai-agents-by-2026-up-from-less-than-5-percent-in-2025 ↑

  3. Gartner, "Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027", press release, 25 June 2025. https://www.gartner.com/en/newsroom/press-releases/2025-06-25-gartner-predicts-over-40-percent-of-agentic-ai-projects-will-be-canceled-by-end-of-2027 ↑

  4. Google, "New ways to create faster with Gemini in Docs, Sheets, Slides and Drive", Google Workspace blog post by Yulie Kwon Kim, 10 March 2026. https://blog.google/products-and-platforms/products/workspace/gemini-workspace-updates-march-2026/ ↑

  5. Anthropic, "The AI-Native SDLC playbook", blog post by Louis Claxton, 21 August 2026. https://claude.com/blog/the-ai-native-sdlc-playbook ↑