Battle for Enterprise AI Control Layer

The real prize is not chat, it is becoming the control layer that sits between a companys data, apps, and AI models. Glean starts from internal search across Slack, Jira, docs, and other systems, then turns that retrieval layer into an assistant and agent product. Writer starts from content generation and brand control, then adds enterprise knowledge and its own Palmyra models. Hebbia starts from deep document analysis, especially dense PDFs and research workflows. These are different entry points into the same budget and the same daily employee workflow.1 2 3 4 5 6

• Glean and Dropbox Dash are betting that model routing matters more than owning the base model. Glean lets admins manage commercial and open source LLMs from providers like OpenAI, Google, Anthropic, and Amazon for Assistant and Agents. Dropbox Dash added a model selector with up to five models. That makes the product valuable even as foundation models keep changing.1 7 8
• Writer is taking the opposite path, bundling the model and the app. Writer says its agents run on Palmyra by default, and enterprise customers can further train outputs on their own examples. That is a tighter stack for regulated content and brand sensitive workflows, where the customer cares less about search breadth and more about repeatable output quality.2 9 10
• The revenue spread shows how early this market still is. By late 2024, Glean was at about $100M ARR, Writer at $47M, and Hebbia at roughly $13M to $17M. That gap suggests broad horizontal knowledge access is winning faster than narrower workflow tools, but Hebbia shows there is still room for specialist products where users spend hours inside documents and need structured answers, not just chat.3 4 5 6

From here, enterprise AI should converge toward products that both find information and act on it. The winners are likely to be the companies that can ingest the most enterprise context, enforce permissions, choose the right model for each task, and turn one off questions into repeatable workflows that save teams real hours every week.1 2 7 8