The world's most recognised AI brand is fighting for its enterprise future on three fronts simultaneously, and struggling on all of them.
"Enterprises require specific and reliable solutions to meet their complex needs," wrote Dovetail co-founder and CEO Benjamin Humphrey. "OpenAI's products are generic and unreliable as non-deterministic horizontal tools."
That structural problem is quietly reshaping how organisations buy AI, and who benefits.
When teams adopt a horizontal AI assistant, every session starts from scratch.
The tool has no memory of last quarter's campaign findings, no awareness of the decisions that shaped last year's product roadmap, and no connection to the structured workflows that turn scattered signals into clear direction.
Foundation models carry vast general knowledge, but they know almost nothing about the specifics that actually drive business decisions.
They cannot retain institutional context across sessions. They cannot replicate the judgment calls that experienced teams have built over years of customer interaction.
The cost is measurable across product, design, customer experience, and sales teams.
Customer feedback analysis carried out on tools with no memory of previous findings means each new project starts blind.
The organisation keeps paying for work that has already been done.
The SaaS companies that understand the context problem are pulling ahead.
"It will almost always be easier to deploy a specific solution that is fit for purpose versus customizing a generic platform to achieve a specific job-to-be-done reliably," Humphrey wrote. "This is also why specific user interfaces will remain superior to a generic chat interface for many applications."
OpenAI's enterprise push is constrained on every front. Amazon, Google, and Microsoft hold decades of infrastructure advantage in cloud services. On the consumer side, Google, Apple, and Meta control the operating systems and browsers through which most people access AI. In enterprise solutions, the middle front, a horizontal toolset, faces the same challenge any generic platform faces: it cannot be all things to all workflows.
The specific beats the general. The CRM that knows your sales pipeline beats the spreadsheet. The customer intelligence platform that knows your customer base beats the chat window that does not.
Dovetail Software centralises customer feedback from every channel: interviews, support tickets, sales calls, surveys, and app reviews. Its AI surfaces patterns, themes, and direction from that data continuously.
Where a generic tool resets with every prompt, Dovetail Software builds a persistent intelligence layer that compounds over time. Product, design, CX, sales, and marketing teams draw from the same body of structured customer knowledge rather than running parallel efforts that produce contradictory conclusions.
AI Channels classify incoming feedback into themes and surface emerging issues automatically, AI Projects convert calls and documents into structured reports, and AI Agents run workflows, flagging risks, generating documentation, alerting teams, all on the same persistent data that grows more valuable as more information enters the system.
Context is not a feature. It is infrastructure.
And it is what generic AI cannot replicate, regardless of how capable its underlying models become.
The early enterprise AI wave prioritised access. Make the tools available, let usage grow, measure volume. The next phase is about strategic deployment: AI systems connected to internal workflows, grounded in company-specific data, producing outcomes that show up in the numbers.
The teams that will lead are those that move from collecting data to extracting meaning, embedding customer intelligence into the decisions that shape the business.
That shift favours platforms built around the job-to-be-done.
Product teams running experience research, CX operations managing voice of customer programmes, revenue and sales teams building intelligence pipelines, marketing teams tracking customer signals, the requirement is the same.
A system that retains context, connects data, and surfaces decisions. Not a chat interface that forgets everything when the session ends.
