Macmillan introduced its workplace AI assistant, NOA, with balloons, banners and doughnuts spelling out its name, according to employees who spoke with WIRED. The celebration presented the tool as a new colleague. Some human assistants, however, found its name—No Ordinary Assistant—offensive, and several staff members later said the tool was not useful or reliable.
A workplace tool with broad ambitions
In an email to staff, NOA was described as an AI chatbot developed by German software company Chaptr, which is backed by Macmillan’s parent company, Holtzbrinck. The email said employees could enter proprietary information in a safe environment without the risk of sharing that data with a third party.
Training sessions reportedly came with a caution: NOA could hallucinate, so its output needed human fact-checking. At the same time, employees say they were encouraged to try it for writing emails, drafting copy, marketing research and idea generation. Staff were also told to paste only a percentage of authors’ manuscripts into the tool.
Users could select a tone for NOA’s responses, with options including “lighthearted,” “spartan,” “pessimistic,” “sarcastic,” and “regretful.” Later in 2024, staff were encouraged to use it for performance-review goals. One email said, “We built a specialized prompt to have NOA write goals for you!”
Adoption and reliability questions
The range of suggested tasks put the assistant into several kinds of work, from routine communications to activities connected to books and employees’ performance. The warning to check its answers acknowledged a limitation, while still leaving staff to decide how much confidence to place in the results.
Employees interviewed by WIRED gave a blunt assessment of their experience. One said, “It was terrible, and no one used it.” Another recalled a Chaptr representative emphasizing that publishers dealt with “content.” The employee responded, “And here I thought we were publishing literature.”
Those accounts point to a gap between the promise of an AI tool and the day-to-day usefulness staff experienced. A system can be available for many tasks without proving helpful in any of them. For publishing workers, the reported concerns also touched on the special care required when dealing with authors’ manuscripts and creative work.
NOA ends as Gemini enters the conversation
In October 2025, Macmillan employees learned in a company-wide email that Chaptr would be sunsetting NOA. Since then, workers say, leadership has been trying to integrate Google Gemini into the publishing process. Some staff experimented with using Gemini to evaluate literary agents’ submission letters; one editor described the resulting evaluations as “useless.”
Macmillan’s vice president of communications and community, Kelly Collins, told WIRED that the company uses Gemini for its “business.” She said there had been “no directive from Macmillan leadership to use Gemini for any part of the publishing process, including to evaluate submission letters from agents.” Collins also said the accounts of NOA’s rollout did not accurately reflect how it was communicated, but did not elaborate when asked which details were inaccurate.
Collins said Macmillan takes a thoughtful approach to AI, with a focus on employee awareness and education and keeping people front and center. The difference between that description and staff accounts underscores how the rollout of workplace technology can be experienced differently by leadership and employees.
A wider debate across publishing
Macmillan’s experience sits within a broader dispute about AI in the book industry. WIRED reports that employees at major publishers described pressure to use AI tools, while Hachette, Macmillan, and Penguin Random House Verlagsgruppe are suing AI companies for copyright infringement. Some bestselling authors said their relationships with publishers had suffered because of AI use.
At Penguin Random House, workers said they were encouraged to use ChatGPT Enterprise in learning sessions, even as the publisher’s German division was suing OpenAI for copyright infringement. Staff described conflicting signals: some said they avoided the tool, while others said managers told them not to use AI for creative work such as publicity copy and cover design. Penguin Random House said its publishing groups and imprints could decide whether AI suited particular workflows, while emphasizing intellectual property and human creativity.
Authors interviewed by WIRED raised concerns about trust, including reported AI-generated marketing copy with fabricated book quotes and an inability to secure a promise that publishers would not use AI on their books. The accounts do not establish one uniform policy across publishers. They do show why questions about manuscript handling, creative judgment and clear communication matter whenever a publisher introduces AI into its work.