The AI Lock-In Trap: Lessons in Contractual Rigor from the ConnexAI Dispute

For enterprise software buyers, the allure of AI-driven efficiency often masks a critical operational risk: the gap between a vendor's marketing promises and the actual performance of the software. When that gap is wide, the exit strategy becomes the most important part of the contract.

As first reported by *The Register*, a legal battle currently unfolding between California-based personal injury firm DK Law and UK-based software provider ConnexAI (registered as Connex One Limited in the UK and Connex One Inc in the US) serves as a stark cautionary tale. The dispute highlights how ambiguous auto-renewal clauses and unverified performance claims can lead to predatory lock-ins and protracted litigation.

### The Performance Gap: Marketing vs. Reality

At the heart of the conflict is a fundamental disagreement over Service Level Agreement (SLA) performance. According to a California court filing, DK Law alleges that ConnexAI sold it a telephone and AI software system—specifically the Athena software—based on functionality that did not exist or perform as represented.

According to *The Register*, the vendor claimed the system would integrate phone and voice services with WhatsApp and SMS, and that the Athena suite included AI Voice, AI Analytics, AI Agent, AI Guru, and Automatic Speech Recognition (ASR). These tools were marketed as a means to provide personalized interactions and 24/7 availability via human-like voice and text communication.

However, DK Law alleges the operational reality was far different. The firm claims the telephone system frequently crashed, resulting in dropped or unanswered calls and consistent misrouting. While the vendor reportedly claimed its ASR "would never miss a word," DK Law alleges that recordings often captured only a few seconds of conversations despite the calls lasting longer. Furthermore, while ConnexAI represented that the software could provide real-time transcripts in both English and Spanish, DK Law alleges that English transcripts were inaccurate and the Spanish transcription failed entirely.

Even after the vendor dispatched US technical services engineers and its UK-based CTO to troubleshoot, DK Law alleges that fixes were only temporary.

### The Auto-Renewal Conflict

As reported by *The Register*, Brendan Haverlock, the CTO of DK Law, informed the vendor that the contract would not be renewed upon its expiration in October 2025. Despite this, the dispute escalated into two separate court cases.

In March 2026, Connex One Inc filed a complaint in a New York court, alleging that DK Law breached its contract by failing to pay invoices. The vendor's filing asserts that the agreement entered in October 2024 "renewed for a subsequent term commencing on October 8, 2025."

Conversely, DK Law filed a complaint in a California court in April 2026, alleging that ConnexAI unlawfully renewed the contract and continued to send invoices and harass employees for payment after the firm had explicitly declined renewal. DK Law further argues that the claim of automatic renewal was not in accordance with the agreement and that the vendor had already breached the contract by failing to deliver services consistent with industry standards.

### Analysis: Avoiding the Predatory Lock-In

This dispute illustrates a common friction point in B2B SaaS: the "auto-renewal" trap. When vendors bake automatic extensions into contracts, they shift the burden of action onto the customer.

For operations leaders, the ConnexAI case suggests several critical audit requirements for AI contracts:

1. **Verifiable SLA Benchmarks:** Contracts should define specific, measurable KPIs for ASR accuracy and system uptime. If these are not met, the customer should have a contractual path to terminate without penalty. 2. **Explicit Opt-Out Windows:** Renewal windows must be clearly defined, and the method for notifying the vendor of non-renewal must be documented and acknowledged in writing. 3. **Performance-Linked Renewals:** The most protective contracts tie the auto-renewal clause to a performance review. If the software fails to meet agreed-upon benchmarks, the auto-renewal should be voided automatically.

### The Vendor's Perspective

ConnexAI has rejected these allegations. Dan Richardson, the general manager of North America for ConnexAI, stated that the claims are part of a wider dispute initiated by Connex One Inc and are "frivolous and without merit." Richardson noted that the allegations were only raised after the vendor had already issued legal proceedings for non-payment.

Ultimately, the resolution of these cases will depend on the specific language of the October 2024 agreement. However, for the broader enterprise market, the lesson is clear: technical due diligence performed during the sales cycle must be mirrored by legal due diligence in the contract.