10-06-2025
22. Writer
Founders: May Habib (CEO), Waseem AlshikhLaunched: 2020Headquarters: San FranciscoFunding: $326 millionValuation: $1.98 billionKey Technologies: Artificial intelligence, deep neural networks/deep learning, generative AI, machine learningIndustry: Enterprise technologyPrevious appearances on Disruptor 50 list: 0
As businesses race to integrate generative AI, Writer has carved out a niche by claiming to have brand-safe, compliant technology that can be tailored for enterprise use. More than 300 major organizations including Accenture, L'Oréal, Mars, Prudential, Salesforce, Vanguard Group, and Uber have adopted its technology, with what the company says is an average nine times return on investment.
Writer offers a generative AI platform that allows companies to build and train AI apps and agents using their own internal data and the startup's large language models (LLMs). They can input documents, style guides, and other organizational knowledge to teach customized AI systems to have "on-brand" outputs within regulatory bounds. Unlike mass market versions of AI tools like OpenAI's public ChatGPT, Writer's systems are designed to integrate with existing workflows via APIs and low-code solutions.
"Companies are struggling to adopt generative AI at scale," Writer co-founder and CEO May Habib told CNBC in an interview last October. "What we do we call a full-stack approach to generative AI. It's the large language models plus the critical tools companies need to customize those models for their data for their workflows."
Last year, Writer launched an LLM called Palmyra X4 meant to compete with enterprise offerings from fellow Disruptors OpenAI and Anthropic, and other market leaders. The average training cost for its new AI model was $700,000, compared with estimates of $4.6 million for a similarly sized OpenAI model, a cost reduction in attributed to using synthetic data created by AI.
Writer's theory is that models that rely on large data sets are hitting their max potential, so synthetic data is necessary to push the AI field forward. The company trained a separate LLM that takes factual data and converts it to data that is structured to train its Palmyra model.
"There's some confusion in the industry about the definition of 'synthetic' data," Writer's co-founder and CTO Waseem Alshikh told CNBC at the time it launched its latest enterprise model in October. "To be clear, we don't train our models on fake or hallucination data, and we don't use a model to generate random data. ... We take real, factual data and convert it to synthetic data that is specifically structured in a clearer and cleaner way for model training," Alshikh said.
The company also released two Palmyra models meant specifically for healthcare and finance companies last year. It also announced Palmyra Vision, a multi-modal LLM focused on analyzing images, extracting handwritten text and classifying objects.
As with all generative AI companies, questions still surround the use cases and return on investment on a widespread scale throughout the economy. While it has been successful in landing big-name clients, it remains to be seen whether those implementations will lead to lasting business growth at a fraction of the cost.
