One startup’s pitch to provide more reliable AI answers: crowdsource the chatbots

Frustrated by expensive enterprise AI contracts that produced unreliable, hallucinated answers, Buyers Edge Platform CEO John Davie incubated a new startup, CollectivIQ. The Boston-based company addresses a critical enterprise pain point: the risk of employees using unsecured AI tools that could train on proprietary data or deliver incorrect information into business presentations. Davie's solution was to build a platform that doesn't force a choice between models but instead leverages the collective intelligence of over 14 major LLMs, including those from OpenAI, Anthropic, Google, and xAI, querying them in parallel to cross-verify information.

The CollectivIQ software operates by sending a single prompt to multiple AI models via their enterprise APIs, then analyzing the overlapping and differing responses to synthesize a more accurate, fused answer. This 'crowdsourcing' approach is designed to mitigate the hallucination problem inherent in single-model queries. From a business and security standpoint, all data is encrypted and deleted post-use, and the company adopts a pay-per-usage model instead of locking customers into long-term commitments. Fully funded by Davie initially, CollectivIQ plans to seek outside capital later this year, positioning itself as a pragmatic tool for companies hesitant to adopt AI due to cost, accuracy, and security concerns.