Ambit AI, an Auckland-based customer service chatbot maker, was placed into liquidation after defaulting on a $467,429 pandemic-era loan, a casualty of the market disruption caused by OpenAI's ChatGPT release in November 2022. The company's collapse illustrates a broader challenge facing AI startups: how to compete when large language models (LLMs), which are AI systems trained on vast amounts of text data, suddenly make specialized products seem obsolete or redundant.

What Happened to Ambit AI's Thriving Business?

Ambit AI had built a promising business before ChatGPT arrived. Co-founder Tim Warren told Tech Insider that the company had reached an annual revenue rate of $2.5 million by 2022, serving major customers including Southern Cross Travel Insurance and Webjet. The firm had developed conversational AI technology tailored specifically for customer service, a niche that seemed defensible and profitable.

But when OpenAI released ChatGPT in late 2022, the market dynamics shifted dramatically. According to the liquidators' initial report, the release of ChatGPT "froze" Ambit's business. Many of Ambit's existing and potential customers began to believe, correctly or not, that they could obtain equivalent products for free or at a much lower cost through general-purpose AI tools. This same phenomenon had already claimed a larger competitor: Soul Machines, another New Zealand-founded customer service chatbot firm, collapsed earlier in 2026 owing $19.6 million.

The company attempted to pivot to a new business model after ChatGPT's initial launch, but the situation worsened when OpenAI released major upgrades to the model. According to Chairman Mark Bregman, the firm "threw in the towel" after these subsequent improvements made it even harder to differentiate their specialized offering.

How Did the Company's Assets Get Divided?

When Ambit AI finally entered liquidation in June 2026, its intellectual property and remaining operations were distributed among several parties. The liquidation process revealed the complex aftermath of a startup's failure and how assets are prioritized under New Zealand law.

• Staff Settlement: Chairman Mark Bregman gave a chunk of Ambit's intellectual property to a group of staff members in lieu of approximately $300,000 owed in salary and holiday entitlements, helping them establish their own AI consultancy called SupaHuman, which now markets itself as a provider of AI solutions for complex, regulated industries.
• Veterinary Sector Acquisition: Another portion of Ambit's intellectual property was sold to Stephen Merchant, a veterinary entrepreneur and former Ambit customer, who formed Ambit AI Powered and took on the original company's three remaining staff members.
• Shareholder Losses: Despite these transactions, shareholders received nothing; Bregman himself lost approximately $500,000 as a shareholder, and other investors including Tim Warren via his Quidnet Ventures also saw their stakes wiped out.

Merchant, a qualified veterinarian who previously co-founded Pet Doctors NZ (which was acquired by Queensland's National Veterinary Care for A$27 million in 2018), has since tripled the size of Ambit AI Powered. He continues to service Ambit's existing clients like Southern Cross Travel Insurance and Webjet, while also incorporating the technology into ClinicWise, an AI-powered virtual assistant designed specifically for veterinary clinics that is now generating global sales.

"We loved Ambit, and its team. Naturally, as a client, we were concerned when we heard Ambit was looking for a new home, and have since done our best to secure and develop the technology, the team and the remaining clients. Tim Warren remains a friend and a valued adviser to us, as one of New Zealand's first AI pioneers and experts," said Stephen Merchant.

Stephen Merchant, Veterinary Entrepreneur and Founder of ClinicWise

Why Couldn't Shareholders Recover Their Investment?

Under New Zealand's Companies Act, shareholders sit at the very bottom of the priority list when a company is liquidated. According to liquidator Adele Hicks, Schedule 7 of the Companies Act places shareholders "at the bottom of the heap" behind staff, the Inland Revenue Department (IRD), secured creditors, unsecured creditors, legal costs, and the cost of the liquidation process itself.

Adele Hicks

In Ambit AI's case, even after the sale of intellectual property to Merchant and the transfer of IP to SupaHuman's founders, there were insufficient funds to clear liabilities including money owed to the IRD. The liquidators' initial report listed only one significant asset: $200,000 in cash in the bank. With the Callaghan Innovation loan of $467,429 and other obligations to settle first, shareholders had no realistic prospect of recovery.

How Can Startups Survive in a World Dominated by Large Language Models?

The collapse of Ambit AI and Soul Machines raises a critical question for AI entrepreneurs: how can specialized startups compete when generalist models like ChatGPT can perform many of the same tasks? Stephen Merchant's experience with ClinicWise offers one potential answer: vertical specialization.

• Vertical Focus: Rather than competing in the broad customer service market, Merchant narrowed ClinicWise's focus to veterinary clinics specifically, where domain expertise and tailored workflows matter more than raw generalist capabilities.
• Integration with Existing Tools: ClinicWise works alongside ezyVet's software for managing veterinary clinics and hospitals, creating a more complete solution than a standalone chatbot could offer.
• Regulatory and Compliance Expertise: SupaHuman, the consultancy formed by Ambit's former staff, positioned itself as a provider of AI solutions for "complex, regulated industries," where general-purpose models often fall short without specialized fine-tuning and domain knowledge.

Merchant explicitly stated that specializing in a vertical was "the way to survive and thrive when ChatGPT and its peers dominated general business". This strategy reflects a broader industry lesson: as large language models become commoditized, the competitive advantage shifts from building the model itself to understanding specific industries deeply enough to apply AI effectively within them.

What Broader Lessons Does This Story Reveal?

Ambit AI's liquidation is not an isolated incident. The Ministry of Business, Innovation and Employment (MBIE) acknowledged for the first time that it is taking a tougher line on delinquent debt from $150 million in loans issued to 456 firms during 2020 and 2021 as pandemic relief. Of those 456 companies, 67 have gone bust, suggesting that many pandemic-era AI and tech startups are struggling to survive in the post-ChatGPT landscape.

The Callaghan Innovation loan program, which provided $150 million in loans at 3 percent interest over 10 years to companies whose research and development programs were negatively affected by COVID-19, was designed to support innovation during a crisis. However, the rapid emergence of powerful generalist AI models has rendered many of those investments obsolete. Callaghan Innovation itself was defunded in last year's budget, with its research role transferred to the new Institute for Advanced Research and its grant and loan responsibilities shifted to MBIE.

The story of Ambit AI serves as a cautionary tale for entrepreneurs and investors: in fast-moving technology markets, even a thriving business with $2.5 million in annual revenue can be disrupted overnight by a paradigm shift. The lesson is not that AI startups cannot succeed, but rather that they must find defensible niches where specialized knowledge, regulatory compliance, or vertical integration create moats that generalist models cannot easily cross.