In the first half of 2025, 64% of all U.S. venture capital funding went to AI startups. That figure alone should make every founder pause and ask: What must I do to stay relevant in a world transformed by AI?
The good news is that many companies can use AI to their advantage. The bad news is that many products simply aren’t suited for this new landscape and will vanish. To avoid becoming a victim of “the prompt,” founders need to get three things right: strategy, product, and organisational efficiency.
Transform your strategy
To reshape your product strategy, it helps to understand what AI truly is, at least in simple terms. AI is another way to input data into a computer to receive an output. We started with punch cards in the 1960s, moved to command lines in the 1980s, graphical interfaces in the 1990s, and now we use natural language and voice.
What’s changed most is the quality of the output. We’ve evolved from simple algorithms to large language models capable of reasoning and problem-solving. But at its core, AI is still an interface for human instructions, and the quality of the results depends on the quality of your input. “Prompt engineering” exists for a reason.
Consider two people using the same AI coding tool: a seasoned engineer with 18 years of experience and a gardener. Their results will differ drastically in user experience, scalability, security, and overall quality. Those who understand both AI prompting and how to build products and communicate with customers will ship faster, automate more, and outperform those relying on outdated SaaS tools.
As AI commoditises much of today’s codebase, you must define what truly sets your business apart. What unique inputs make your model work? Inputs that are hard or impossible to replicate with prompts? These could include regulatory licenses, distribution channels, contracts, relationships, skilled labour, or even company culture. Identifying this will make your business more defensible and less attached to existing code or features.
If you can’t articulate such an advantage, your product might just be a system of record, vulnerable to being replaced by a better-prompted alternative.
Transforming your product
“Coding is fast. Deciding what to code is slow.” – Andrew Ng
Once you’ve clarified your strategic advantage, you can reimagine your product. Treat AI as another channel, like web or mobile, rather than a revolution in itself. Ask which user experiences make more sense as conversational (text or voice) rather than clicks and forms.
But don’t just build an AI agent for the sake of appearing modern. Having one is already becoming table stakes, like having a mobile app a decade ago. Instead, focus on where your existing assets and capabilities allow you to build something truly differentiated.
AI features shouldn’t exist to impress investors or prove you’re “using AI since before 1984.” They must solve concrete customer problems. The opportunity is huge: over 90% of enterprises plan to adopt AI, but only 5% of trials succeed, according to MIT. The main reason? Most AI tools don’t integrate into existing workflows. They ignore the unglamorous but necessary foundations and workflows that enterprises depend on, like security, policies, and audit trails.
Once you know where your product fits, focus on how you build it. This isn’t about choosing the right code-generation tool (already used by 84% of developers). It’s about redefining product leadership. The role of the classical Product Manager planning for 12 months ahead in a traditional company is gone.
The new Product Leader merges customer insight, user experience, design, and architecture. Instead of delegating, they prototype directly through “vibe coding”, rapidly experimenting with features that accelerate organisational learning. If a PM can’t effectively prompt AI, can they truly guide engineers and designers?
Transforming your organisation
A company’s product is the sum of its teams, services, and processes. You can’t reinvent the product without transforming the organisation itself, and that’s the hardest part.
Native AI companies have a clear edge here. Lean and fast, they can achieve $500M valuations with teams of just 20–30 people in under two years. While off-the-shelf AI tools can improve individual productivity, MIT found they deliver meaningful business impact only about 5% of the time, despite tens of billions invested. The biggest obstacle isn’t the tech; it’s people and, more specifically, managers, reluctant to change.
The low-hanging fruit (5%)
Adopting AI tools like ChatGPT, Cursor, or Copilot may boost individual performance but rarely transform organisations. Around 80% of companies have tested such tools, and 40% use them regularly, yet few see direct P&L impact.
Why? Because every company can adopt them. Real advantage comes from efforts that require leadership commitment, not just tool adoption: systematically rethinking workflows, retraining teams, and changing processes.
The next step (10%)
Some off-the-shelf AI products can significantly enhance performance in specific departments. Take Intercom’s Fin, for example. It can handle 50–60% of customer support autonomously and up to 90% when integrated with company data. It improves help articles, creates new ones, and maintains smooth handoffs between AI and human agents. With robust reporting and analytics, it turns support teams into AI copilots rather than watchdogs.
The promised land (85%)
Reaching full transformation means deeply automating how work gets done. Tools like n8n, Glean, Lovable, and Make allow companies to build internal automation that minimises human error and accelerates workflows. But true transformation requires more than tool adoption. It demands vision, coordination, and cultural change.
Organisational complexity grows with headcount. Processes slow, knowledge becomes siloed, and information flow suffers. An incumbent company that is not AI-native will require significantly more time and resources to adopt AI and transform its processes, people, and tools into a truly native AI company that has managed to achieve exceptional operational efficiency with $1M+ ARR per FTE.
To illustrate the challenge, consider the IT Core Complexity model, which measures transformation difficulty based on communication channels, role diversity, and workload. The formula is:
Complexity = (Communication Channels + Role Complexity) × Stress Factor
According to Conway’s Law, system design mirrors an organisation’s communication structure. A 10-person company has 45 unique communication channels; 100 people generate 4,950; 500 employees create 124,750. Each additional link multiplies friction. Even if AI halves transformation time, a 100-person company still needs about a year to fully adapt—assuming this is a top priority across the organisation.
Below is a breakdown of the inputs and timeline to transform an organisation based on its IT Core complexity. We can even assume that AI will take half the time, but still, we are looking at an average of 12 months to transform a business with just 100 employees if this is a full-time initiative across the board.
| Company size | Unique connections | Roles to transform | Initiatives per employee | Time to transform |
| 10 employees | 45 | 3 | 3 | 3 – 6 months |
| 100 employees | 4,950 | 10 | 2.5 | 12 – 18 months |
| 500 employees | 124,750 | 15 | 2 | 2.5 – 4 years |
The math makes it clear: smaller teams can adapt faster. Agility itself becomes a competitive advantage in a rapidly shifting market. Yet even AI-native startups risk falling into the same bureaucracy trap if they don’t consciously maintain simplicity.
Historically, large-scale technological transformations take three to five years to show up in company P&Ls. The AI revolution will be no different. Founders must now decide: will you lead, follow, or do nothing and risk becoming irrelevant?
