Writer, an AI framework capable of generating written content for marketing teams, announced today that it has secured $21 million in series A funding, in a package led by venture capital company Insight Partners.
Contributing funders include a variety of high-level investors and CEOs, including the Todd Goldberg Angel Fund; New Normal Fund founder Allison Pickens; Adobe Creative Cloud Product Officer and Executive Vice President Scott Belsky; Head of Global Revenue & Business Development at Asana Oliver Jay; co-founder of Superhuman and LiveRamp Vivek Sodera; and Google’s AI-centric venture fund, Gradient Ventures, among many others.
The Writer platform is based in San Francisco, and has raised $26 million USD to date, with $15 million of seed funding garnered in August of 2020. The latest round is reported by BusinessWire to be intended for engineering, GTM hiring and product development.
The Writer platform offers a slew of use cases for AI-assisted language generation in the enterprise, including content strategy, marketing, editorial, documentation generation, learning and development, support services, HR and Operations and IT.
A sustained period of growth has seen the company gain new customers such as Vistaprint, Handshake, Deloitte, United Healthcare and InVision.
Bringing Structure to Copy
Describing writing as ‘the last unstructured business process’, co-founder and CEO May Habib says (not inaccurately!) that ‘about 75% of the time spent writing is actually time spent thinking, searching for information, or looking for something someone else has written before’, and believes that the company’s diverse offerings are set to change this, or at least automate it.
The core technologies of the company center on aiding those who must produce content, from writers through to managers and team leaders, rather than achieving the more ambitious goal of getting AI to identify and write long-form content unaided.
The platform offers free grammar checks as well as tools to generate style guides and to compare contributed content to company style.
The platform offers a single and unified lexical and guideline resource for companies looking to scale content production, with a range of tools variously aimed at technical and marketing writers, and group-centered content management as well as single writers.
Also available are Content Templates that can lay out the basic structure of an article or post as a framework for the writer or writers to flesh out. Further, line-by-line feedback on content can provide detailed insights into how submitted copy squares up against the custom-defined styles and standards of the company.
NLP and Core Technologies
In January of 2020 the company open-sourced their custom-built FitBERT Python library. FitBERT stands for ‘Fill in the Blanks, BERT’, and operates more as a kind of lexical polyfill than a full-fledged encoder/decoder-style NLP generation engine.
FitBERT encapsulates Writer’s halfway-house business proposition quite well, by adapting the occasionally overstated abilities of Google’s BERT model to tasks that it is more likely to achieve correctly and consistently, such as spell-checking, grammar analysis, and pattern-matching as it relates to style analysis.
The code can also be experimented with live in Colab.
Plugged In Analytics
Though the array of network-facing plugins offers the potential for the company to re-use client content to refine its machine learning models, the terms of service preclude this, and the company states:
‘What you write will never make it into our machine learning models. Your data is analyzed in your browser, so what you write in your online tools stays there.’
The Writer platform also offers analytics that can discern which categories of errors are recurring in teams, as well as reporting on the general install base and usage of the product in a company.
Automating Linguistic Conformity
Writers with idiosyncratic styles are likely to have their approach smoothed over by Writer’s weekly reports, which indicate a number of possible areas for attention, including ‘wordiness’, ‘unnecessary commas’ (which may upset fans of the Oxford comma), and use of the passive voice. Naturally, all these suggestions will originate from styles already defined by the company using these services.
May Habib says:
“Most teams don’t have the resources to ensure strong, consistent writing across every single person who works at a company, so we provide a seamless way to help everyone write well, write fast, and be on-brand. This funding helps us achieve our vision of great writing for everyone faster.”