Writer implements large, in-house developed language models to enable corporate copying

There’s a lot of buzz these days about how generative AIs like ChatGPT and Bard will revolutionize various aspects of the web, but companies targeting narrower industries are already seeing success. Writer is one of them and just announced a new trio of great language templates to power your business copy assistant.

The company allows customers to customize these templates to their own content and style guidelines, from which the AI ​​can write, help draft or edit text to meet internal standards. Not only can Writer’s new templates detect typos and recommend a preferred word, but they can self-evaluate style and write content, even doing a little fact-checking when they’re done.

But the real kicker is that everything from tuning to housing can be done in-house, at least as far as the two smaller models in the Palmyra range are concerned.

“No company leader wants their data fed into anyone else’s base model, including ours,” CEO May Habib said in a press release. “We offer customers all the benefits of the AI ​​application layer without the risk of other AI applications and business models. Business leaders want to invest in solutions that essentially earn them their own LLM.”

Palymra is available in three sizes: 128 million, 5 billion, and 20 billion parameters for Small, Base, and Long, respectively. They are trained in business writing and marketing, not Reddit and Project Gutenberg posts, so there are fewer surprises in the beginning. Then you fill your mouth with annual accounts, finances, blog posts, etc. from the last ten years. to make it yours. (This and any derived data is not filtered back to Writer, just to be clear.)

Having written my share of business and marketing copy, I can tell you that this is not the most exciting app. But what it lacks in excitement it makes up for in practice: Companies have to do a lot of this kind of writing and editing, and they tend to pay for it. Writer is already plugged into many development and productivity suites, so it doesn’t cause much friction.

The business model is similar to other generative AI companies: set up and customize everything for free, and pay a penny for a thousand tokens, which will get you about 750 words. (This article has just over 500, for quick reference.)

Alternatively, you can clean the Small or Base models yourself for free if you have the computer.

A few dozen companies have been using the models since late last year, and we’ve never heard of serious problems like the one on the first day of Microsoft and Google’s attempts to popularize generative AI…so that’s a good sign. . That’s the achievement I was talking about earlier. While ChatGPT is certainly impressive as a generalist or amateur AI, it’s hard to say what it can really be used for. In the next two years, we’ll see more specific games like Writers, as Microsoft and Google kick their latest toys.