Many companies now claim to be AI-native. A check on the Wayback Machine can reveal when they started using AI terminology. If it is only after ChatGPT's launch, their claims warrant deeper scrutiny as they might be narrative-driven rather than core to the business, a key diligence check.
During M&A, don't rely on key personnel to explain your AI advantage. Instead, create a separate document presenting the 'game-changer' with a single, compelling chart. Provide the raw backup data to prove the chart is not hallucinated, ensuring a cohesive and trustworthy story that stands on its own.
CEOs expect AI to massively increase feature output, but data from tools like Jellyfish shows AI-generated code often requires much longer review cycles. This can slow a team's overall momentum despite the high volume of code produced, creating a disconnect between executive expectation and engineering reality.
Many businesses overlook their most valuable existing data assets. Years of unstructured data, like support tickets detailing integration issues and customer problems, are invaluable for training specialized AI models. This 'boring' data can become a key source of competitive advantage when activated with an LLM.
A few months ago, the fear was AI replacing SaaS businesses. Now, the pressing issue is managing massive AI bills. This has elevated 'token economics'—optimizing costs by using different models for different tasks (model routing)—from an advanced technique to a non-negotiable, table-stakes practice for any serious AI implementation.
AI tools empower non-technical CTOs—those who primarily manage vendor contracts—to believe they can direct engineering. They push for rapid feature development without understanding engineering fundamentals like test coverage, creating massive technical debt. This is akin to giving a toddler a handgun.
Over-reliance on AI for coding can lead to engineers simply approving auto-generated code without deep understanding. This 'cognitive atrophy' erodes fundamental skills, creating significant long-term risk for the organization as nobody is truly responsible for the codebase's integrity or logic.
Companies with outdated codebases were often too risky to acquire due to modernization costs. AI-powered code translation tools now make it feasible to refactor legacy systems into modern languages relatively quickly, opening up a new category of M&A and roll-up targets that were previously untouchable.
High-performing teams are creating small 'pods' with a product manager (business context), a designer (UI/UX), and an engineer (technical execution) who work together in shared AI coding sessions. This collaborative model ensures features are viable, usable, and well-built from the start, breaking down traditional silos.
AI prototyping allows product managers to build out every feature of an idea cheaply. This process of externalizing the full scope often reveals which features are unnecessary. This rapid iteration helps them 'detox' from initial over-scoping and arrive at the core value proposition before involving engineering.
The most valuable AI assets are not models, but proprietary data from years of solving domain-specific problems. This 'scar tissue'—like knowledge from undocumented APIs or complex integrations—is painful to acquire and impossible for competitors to replicate quickly, creating a durable competitive moat.
Companies are pushing for more AI-generated code to cut costs, but this code is often not fully understood by engineers. This creates significant security vulnerabilities that advanced AI models will inevitably exploit, potentially destroying smaller companies that fail to maintain a strong security posture and rigorous engineering standards.