Person
1 episodes · 15 quoted claims
May Habib says that in 2023 Writer described the dominant design for generative AI as language models plus retrieval, guardrails and an AI studio.
“in 2023, we said dominant design and generative AI looked like LLMs and retrieval and guardrails and an AI studio”
May Habib says Uber, Salesforce, Accenture and NVIDIA use Writer internally.
“customers like Uber and Salesforce and Accenture and NVIDIA are using Writer internally”
Uber's support team uses Writer agents to classify Jira and Slack requests, assess urgency and generate updated documentation content, with humans approving publication.
“they use Writer agents to classify Jira and Slack requests, they assess urgency, and they generate updated responses and content”
At the cybersecurity company Commvault, Writer's AI agents automatically pull RFP files from Salesforce as the start of a proposal workflow.
“AI agents from Writer were automatically pulling RFP files from Salesforce.”
May Habib says Writer builds 'blueprints' for agents rather than fixed workflows because workflows are too deterministic for goal-oriented agents.
“in Writer, we build blueprints, we don't build workflows, because that's that's very deterministic”
Writer has built lifecycle management tools that let customers retire agents that cost too much relative to their value.
“we've also built just lifecycle management for agentic. So how do you deprecate them if they cost too much relative to the value”
May Habib says conventional search is keyword-based while language models prefer comprehensive paragraphs and sentences, which shapes generative engine optimization.
“So much of conventional search is keywords based, but LLMs really like nice, comprehensive paragraphs and sentences”
May Habib says Writer's top thirty or so customers have a net revenue retention of 217%, meaning they are roughly tripling their spend.
“our top thirty-ish customers are at 217% NRR. You know, this is customers that are tripling spend with us”
May Habib says Writer's Palmyra X4 and X5 models are the main models behind its agent workloads.
“our Palmyra X four and X five are are the workhorses of agentic workloads”
May Habib says Writer's Palmyra models have more than 500 billion parameters but are designed to be efficient.
“They're, you know, 500 billion plus parameter uh models, but we have architected the transformers to be highly efficient”
May Habib says Writer does not train its models on customer data and that they are trained on synthetic data.
“We don't use their data for training, or the models are actually trained on completely synthetic data.”
May Habib says Adobe Firefly is a partner and that 95% of Writer customers' workloads run on the Palmyra models, with Gemini used for some image generation.
“Firefly is a big partner of ours, but, uh, 95% of the workloads are on the Palmyra models.”
May Habib says not training on customer data has become standard, and that OpenAI also no longer does so after a certain level of customer spend.
“it's become table stakes to not train models on folks's data, right? OpenAI's not doing that either anymore after a certain level of spend”
May Habib says Writer's most successful customers roll AI out to large groups of clinicians in healthcare and use it to grow their business.
“The most successful customers, right, that that we have seen are going to groups of 300, 400, 500 in healthcare, nurses, doctors, clinicians”
May Habib says Writer's workforce is more than 40% women and balanced across teams.
“we're more than 40% women, um, and pretty even across all teams”