Addressing the Challenges of Enterprise AI Adoption
At the recent VB Transform event, May Habib, the CEO of Writer, discussed the latest advancements in the company’s AI platform and shed light on the obstacles hindering enterprise AI adoption. Writer, an enterprise AI platform, has introduced significant improvements to its artificial intelligence chat applications, including advanced graph-based retrieval-augmented generation (RAG) and new tools for AI transparency.
The upgraded chat apps, set to go live soon, boast enhanced data processing capabilities. They can now analyze up to 10 million words of company-specific information, benefiting users of the “Ask Writer” application and developers utilizing the AI Studio platform for custom solutions.
Habib acknowledged the challenges facing enterprise AI adoption, stating that many companies are struggling to demonstrate significant progress in their AI initiatives. She emphasized three main obstacles: low accuracy, inefficiency, and poor adoption rates. According to a survey of 500 AI executives, only 17% rated their AI applications as “good or better,” indicating that the majority of enterprise AI efforts are falling short of expectations.
To address these challenges, Writer has developed a “full stack generative AI” approach. Central to this approach is their graph-based RAG technology, which maps semantic relationships between data points for more targeted information retrieval. Habib showcased how the platform’s graph-based RAG system interprets complex queries and provides clear reasoning for its outputs, enabling non-technical staff to effectively work with AI and improve accuracy, efficiency, and adoption in real business settings.
In addition to graph-based RAG, Writer has introduced the “thought process” tool, offering transparency into AI decision-making. This tool breaks down complex queries into sub-questions that the AI assumes users are asking. Furthermore, specialized “modes” have been implemented to streamline user experience and improve output quality.
Habib also highlighted the need for more user-friendly AI tools in enterprise settings, referencing the low adoption rate of Microsoft’s Copilot. She mentioned that if 50% of employees use Copilot once a week to summarize an email, it would be considered in the top decile of adoption. This statistic underscores the importance of developing AI tools that are accessible and easy to use for a wider range of users.
In conclusion, Writer’s latest updates address the challenges of enterprise AI adoption by enhancing data processing capabilities, improving accuracy and efficiency, and providing transparency into AI decision-making. These advancements pave the way for more effective and user-friendly AI applications in enterprise settings.