AI Vendor ConnexAI Faces US Lawsuit Over Software That Allegedly Didn't Work and a Contract Renewal Nobody Asked For
A California personal injury law firm is suing UK-based AI and telecoms vendor ConnexAI, claiming the company sold it software that barely functioned and then kept billing it after the contract ended. ConnexAI says the allegations are rubbish. Both sides are now in court.
DK Law, based in California, filed a complaint in April in a US federal court alleging that ConnexAI makes promises to customers about "functionality that does not exist or perform as represented." That's a fairly damning opening line, and the filing doesn't get much kinder from there.
The background is a year-long contract for ConnexAI's telephone system and its Athena AI platform. The sales pitch was substantial: phone and voice services integrated with SMS and WhatsApp, plus AI features covering sentiment analysis, entity recognition, call transcripts, interaction clustering, and automatic speech recognition. Crucially, ConnexAI apparently told DK Law the transcription worked in Spanish. That detail matters because DK Law handles personal injury cases in California, where Spanish-speaking clients are not exactly a niche audience.
According to DK Law, the reality was considerably less impressive. The phone system went down regularly. Calls were dropped, missed, or routed to the wrong places. WhatsApp integration was effectively useless. The speech recognition, despite being pitched as something that would "never miss a word," apparently missed quite a lot. Spot-checking call recordings revealed that some captured only a few seconds of what were clearly much longer conversations.
The Spanish transcription? Didn't work at all, DK Law claims. English transcripts were produced inconsistently and, when they did appear, were inaccurate.
Support tickets were raised. ConnexAI reportedly sent engineers over from the US and flew in the UK-based CTO to investigate. The problems were occasionally patched up, then came back. No permanent fix ever materialised, the complaint says.
By the time the contract was approaching its October 2025 expiry, DK Law's CTO Brendan Haverlock had apparently seen enough and told ConnexAI the firm would not be renewing. That should have been the end of it.
It wasn't. ConnexAI kept sending invoices. When DK Law didn't pay, ConnexAI's US entity filed a breach of contract claim in New York in March 2026, arguing the contract had automatically renewed for a further term starting October 2025.
DK Law's response to that was to file in California, arguing two things: first, that ConnexAI had already breached the contract through poor performance; and second, that New York state law requires vendors to give written notice before an automatic renewal kicks in for telecoms and software agreements. ConnexAI, DK Law alleges, never sent that notice.
ConnexAI's North America general manager Dan Richardson offered a fairly tight-lipped statement, noting the claim was "one part of a wider dispute" that ConnexAI itself initiated, that the allegations are rejected, and that they only surfaced after ConnexAI launched its own proceedings. That last point is worth noting, but it's also not unusual for defendants to file counterclaims.
ConnexAI also questioned why its UK parent company, Connex One Limited, was named in the California action given the original contract was with the US subsidiary. DK Law's answer is that the UK entity ran the sales process, controlled implementation and support staff, and approved the post-contract renewal conduct. The US company, DK Law argues, is wholly owned by the UK parent and effectively acted as its agent.
For now, the California case is on hold. A judge ruled last week to stay the proceedings pending the outcome of the New York case. So the New York breach-of-contract claim goes first, and the California fraud and performance allegations sit in a queue behind it.
It is a mess that will likely take a while to untangle. What it illustrates rather neatly, though, is a pattern that will be familiar to anyone who has watched enterprise AI software deals go wrong: ambitious sales claims, underdelivered functionality, and a dispute about whether the customer is obligated to keep paying for something that didn't do what it said on the tin.