Earlier this month, author founder and CEO May Habib took the stage at the HumanX conference in San Francisco to discuss with Bloomberg's Natasha Mascarenhas why many companies' AI strategies are failing. In a situation where 95 percent of AI production pilots fail to deliver returns, Mlobi is involved in helping Fortune 500 companies build AI agents and automated workflows that integrate with existing systems. Its offerings include Writer Agent, a multi-step workflow platform, and Palmyra, a family of large-scale programming languages designed for business use. The San Francisco-based startup last raised $200 million in November 2024 at a valuation of $1.9 billion. If gold rushers reward shovel salesmen, the AI race rewards practicality.
Habib and his co-founder and CTO, Waseem Alshikh, met in Dubai in 2011 after reaching out to Twitter about his work in machine translation. The connection quickly evolved into a partnership, leading to the launch of Qordoba, a startup focused on AI-powered content localization. In 2020, they include Writer.
“Honestly, I think it's up there with one of the five best things, days, times, lucky breaks I've ever had, meeting my founder and deciding to do this with him,” Habib told the Observer.
Habib was born in a village on the Lebanon-Syria border during the Lebanese Civil War. His family fled to Canada in 1990. As the eldest of eight children and the only one who spoke English, she often served as an interpreter, a role that later developed her interest in linguistics. He graduated from Harvard in 2007 with a degree in economics and a minor in Oriental languages, eventually parlaying that experience into work in natural language processing and machine learning, including founding Cordoba.
At Author, the co-founders maintain a complementary dynamic: Habib represents the voice of the customer, while Alshikh represents the technology. The company initially positioned itself as an “AI writing assistant for professional users,” helping organizations increase communication. In early 2025, it turned to AI agents following advances in model thinking.
“Our focus has always been on multi-step, real-world workflows, and now we have models that can implement these workflows with minimal errors, minimal human intervention, and because of the way we've implemented our product, in a loud, transparent, [and] the decline that business needs,” said Habib.
The author declined to share revenue figures but said it has more than 300 business clients who refer more than 15,000 agents. Clients include Ally Financial, AstraZeneca, Cigna, Clorox, Comcast, Franklin Templeton, Keurig Dr Pepper, Lennar, Marriott, Mars, Uber and Vanguard.
One of its most notable customers is Salesforce, where more than 3,000 employees use Writer to query and generate content from internal data. The company also trained 50 internal “champions” to build custom agents and workflows using code-free tools. Airbnb is another client, using Writer to generate content for 37,000 web pages during its Experience rollout.
While many AI startups emerged after the launch of ChatGPT in late 2022, Mlobi preceded the boom and saw it as a competition.
“You know, our pencils have been sharpened. We've sharpened our delivery. Our great speed, and the speed of our product, we owe 100 percent to that great competition that's coming on the scene,” said Habib.
“The author was founded before the ChatGPT moment changed the market. They were completely focused on the business from the beginning,” Shashi Bellamkonda, chief research director at Info-Tech Research Group, told the Observer. “What protects them is what they've already built: compliance, deep customization, and a family of proprietary models that give customers full control over their data and results.”
Asked why many companies are struggling to incorporate AI, Habib pointed to a focus on incremental improvements rather than structural change.
“I think that taking 54 handoffs in a program and making it 27, it's not only uninteresting, it's not working. “You need a new model of how work is done…
Companies stuck with legacy processes are at risk of falling behind. As an AI-first company, Author emphasizes speed and accuracy. “Maybe we do things in a day that take other companies a month. We have to make a decision, because if we don't, we are pressured, and that is a big part of the working rhythm,” said Habib.
Security remains a major concern for AI agents. A Meta researcher recently shared an incident where OpenClaw ignored instructions and started deleting his email inbox. At Amazon Web Services, an agent named Kiro “deleted and rebuilt” the production environment, causing a 13-hour outage. These situations highlight the dangers of insufficient vigilance.
The author addresses this with layered controls, including a killswitch at the agent, tool, connector and user levels, allowing companies to block unwanted actions. These protections are enabled by its cloud-based architecture.
“It's easy to allow an agent system on your desktop to do whatever the f*** wants. It's very difficult to spend years building a cloud-based solution where connector by connector, you have the ability to determine which user has the skills to access which type of action,” said Habib. “Being able to identify that [an] agent workflow… That's really our secret sauce.”
That focus on managing and controlling the quality of the business drives the writer's momentum. Even amid concerns about the AI bubble, the company is betting on a solid ground: make AI-first really work. Bright adoption cycles may be ending, but businesses will continue to invest in tools that make core operations more efficient.
