By 
By Tracy Yekaghe
Startup Writer has secured $100 million in series B funding, establishing a valuation between $500 million and $750 million, as revealed by the company. Writer’s extensive language models generate a wide array of content, including incident reports, emails, product descriptions, and executive summaries, positioning it directly against OpenAI’s ChatGPT Enterprise, introduced just last month, as well as other rapidly expanding unicorns like Typeface.
In the midst of a competitive generative AI landscape, Writer’s CEO and co-founder, May Habib, informed Forbes that a portion of enterprise clients have transitioned from Azure OpenAI to Writer due to concerns about the output quality from ChatGPT. The startup has seen a remarkable revenue surge, growing tenfold in the past two years and quadrupling since the beginning of this current year.
“We are seeing a lot of folks reach out as they’re kind of stuck in proof-of-concept purgatory,” she said. “We’ve seen a few customers where they haven’t been able to take use cases to production because the generations weren’t high quality enough.”
Writer, established in the year 2020 and highlighted on Forbes’ AI 50 list earlier in this year, highlights its roster of 150 enterprise clients, including Uber, Spotify, Vanguard, Samsung, Accenture, and L’Oreal. The funding round was spearheaded by Iconiq Growth, joined by WndrCo, Balderton Capital, Insight Partners, and Aspect Ventures. This funding round has boosted the startup’s overall funding to a total of $126 million.
Writer presents a range of 14 models of varying scales, spanning from 128 million parameters to 43 billion parameters—significantly smaller compared to OpenAI’s GPT-4, which is reported to possess approximately 1 trillion parameters. These models, named Palmyra, underwent training using publicly available data extracted from diverse sources like web pages, books, Wikipedia, Github, and transcribed video content from YouTube. The publicly sourced data is meticulously filtered to eliminate copyrighted material, as mentioned by Habib. Every enterprise receives a distinct fine-tuned variant of the model, customized using company-specific confidential data such as financial reports and marketing content. Moreover, these models adhere to a majority of privacy and security regulations, including HIPAA and Europe’s GDPR.
In the latest development this July, Writer introduced PalmyraMed, a model specifically trained on publicly available medical datasets such as PubMedQA, encompassing both questions and answers along with articles, tailored for applications in the healthcare sector. Habib mentioned that Writer’s more compact fine-tuned models are trained at a mere fraction of the cost compared to larger language models. These models are designed to efficiently execute specific tasks at accelerated speeds.
Habib pointed out, “The difficulty for enterprises lies in fine-tuning a large language model (LLM) that’s proficient in a wide range of tasks using their extensive dataset and crafting a model capable of addressing any problem. The crucial aspect is the final stage of fine-tuning, data preparation, and seamless integration into existing workflows where enterprises often require assistance.”
Habib emphasized that a pivotal advantage of Writer is its seamless incorporation with the software employees commonly utilize, such as Salesforce and Adobe. She added that the application can be seamlessly integrated into various workspaces like Google Chrome, Figma, Google Docs, Canva, Microsoft Word, and Outlook by installing Writer’s plugin or extension. These integrations automatically annotate the content with suggestions, enhancing productivity. She also mentioned that the initial appeal of chat functionality has diminished.
