From heights to the brink: Silicon Valley, once a technology leader, now confronted by new AI forces

According to Sumeet Vaidya, CEO of Crafting, reported by Fortune, the current shift in Silicon Valley represents a disruption on many fronts. Historically, the region’s best startups stayed ahead of emerging trends by several steps, building durable businesses by anticipating what people would want as new waves took hold.

Pressure mounts on tech giants as user demands change

Vaidya argues that Silicon Valley is now lagging behind the societal norms it helped set. He says large incumbents are focused on protecting closed business models, which reduces how vigorously they innovate. As models and features update rapidly, the pressure on tech giants is increasing.

He also contends that the “manuals of power” that once guided society through the early web, mobile, and cloud eras are now obsolete. In his view, tech leaders must yield to what users want, rather than forcing users to remain within their ecosystems.

AI assistants need orchestration across ecosystems

Vaidya says the future of AI assistants requires orchestration and coordination layers that do not depend on a user’s phone, a company’s cloud provider, or the specific model a user prefers. He adds that this is feasible today, but that AI leaders are “in denial” and are determined to keep users inside their own ecosystems.

He further argues that privacy and security concerns are being deprioritized, with some believing they can tailor agents using large data stores such as emails, iMessage conversations, meeting notes, and other daily software interactions.

From code generation to testing bottlenecks

Vaidya points to a practical bottleneck at the center of Silicon Valley’s AI push: in the rush to demonstrate that AI can generate high-quality code, pioneers are now overwhelmed by large volumes of AI-generated code that are difficult to verify.

He says the bottlenecks have shifted from writing code to testing it, and even rechecking AI agents’ test suites. He cites an AWS outage caused by AI as a “painfully highlighted” example of the risks involved.

As a result, engineers are spending more time rechecking what AI has written and waiting for test processes to complete. Vaidya says this is raising questions across engineering organizations, from CEOs to junior backend engineers, about whether teams are truly more productive.

“Write one and know ten” becomes the standard users expect

Vaidya argues that AI tools are only as strong as their capabilities, and that the ideal assistant would be usable anywhere and anytime. He describes an approach used by pioneers: buying up Mac Minis and loading the software they use onto them, aiming for AI that can draft messages, access email, review meeting decisions, monitor trade progress, and understand branding rules.

He says no company has succeeded at building this kind of all-encompassing capability, but that the direction is clear: AI models are increasingly becoming commodities, and giants must prepare for a multi-model, multi-ecosystem future to close the innovation gap and regain leadership.

Examples of emerging orchestration and agent ecosystems

Vaidya highlights NanoClaw as an example of how AI may look in the future, and notes the rapid adoption of tools like Town for individuals seeking a 24/7 efficient assistant.

He also describes how Crafting is being used by enterprise tech teams to connect data from many disparate sources to build agents capable of deploying products like software engineers. He adds that lower-level tools such as Zapier and Gumloop are attracting non-technical users who are building personal agent orchestration systems.

Impact and outlook for the next phase of competition

Vaidya concludes that people do not oppose AI if it can do what they want. He says the benefits are expected to bring AI into everyday life for the masses and into large-scale businesses.

While he says it is unclear which companies will endure the competition as the AI era peaks, he argues that startups may ultimately become the “glue” binding ecosystems together. He says the giants that have shaped how the world experiences technology must respond to social demands by building AI ecosystems people truly want, or risk losing everything.

Source: Fortune