Nearly three years after ChatGPT’s public debut, very few business leaders need convincing that their organizations should aggressively adopt AI. But even teams that are committed to embracing the technology are having trouble developing practical solutions that create real business value, says Mike Morris, chief technology officer and chief AI officer at Effectual.
“We’ve been working with a very large, very technologically advanced customer in the defense industry,” Morris says. “They had a team that was four or five months in, and they couldn’t even get a proof-of-concept [PoC] up because they were so overwhelmed by the rate of change and how many options there were. This is certainly not unique. This is something we’ve heard from many organizations that have tried to do this on their own.”
The problem, Morris says, lies not in a lack of effort or ambition, but rather a lack of flexibility. Many organizations, he notes, have locked themselves into rigid platforms, proprietary tooling, or one-off experiments that are difficult to scale. By breaking organizations free from these constraints, Effectual has seen over a 90% success rate helping customers move their AI projects from PoC into production.
Most critically, Morris advises leaders to select an AI platform that provides access to multiple large language models (LLMs). The space is moving so quickly that these models are adding new features seemingly weekly, and there is no guarantee that today’s leader will still be ahead tomorrow. “If a model provider announces a new feature, you need to have same-day access to that feature on your AI platform,” Morris says.
Often, organizations will develop a simple AI PoC using methods like retrieval augmented generation (RAG) for a very small subset of their data to improve the accuracy of generated outputs. Some organizations have successfully used this approach to provide answers to sales and field teams based on internal documentation, or to rapidly summarize call logs and increase productivity for customer service teams.
However, many organizations have difficulty with these use cases. If AI platforms aren’t flexible enough to integrate new capabilities quickly and securely, even promising use cases can stall.
“If you get locked in, you’re not going to be able to take advantage of the latest innovations,” Morris says. “Your competitors are going to avoid that, and you should do the same. Don’t go with an ultra-proprietary platform.
Effectual delivers AI accelerators within customers’ Amazon Web Services (AWS) environments, using cloud-native services to ensure rapid deployment and a zero-trust security approach to maintain data privacy and confidentiality. The scale of AWS allows organizations to innovate without resource constraints.
Open-source solutions, Morris stresses, can help organizations maintain the flexibility they need to move quickly during a time of great change. He also emphasizes the importance of partners in helping keep enterprise IT initiatives on track, as well as preventing internal leaders from getting bogged down in analysis and failing to act.
“It’s very easy to get overwhelmed with how many options there are in the space, and with the pace of innovation,” Morris says. “An experienced partner can jump-start things and get your organization moving in the right direction.”
Request your complementary consultation with an AI Architect.