Turning Decades of Agency Know-How Into an AI Product
Thread AI Editorial · October 2, 2026
A global marketing and advertising agency is made up of several specialized business units, each with its own teams, tools, and ways of working. Over decades and thousands of client engagements, the agency built up a clear sense of how its work should be done, but that know-how didn’t reach everyone who needed it.
The agency built its own branded product on Lemma that puts its expertise to work for every employee, from a new hire’s first week to the work that goes out to clients. Work across every business unit can now reflect the same standards.
What Lemma delivers
1 in 3
Employees using it within seven weeks, with no mandate to adopt it.
2,600queries
Delivered in the first seven weeks after rollout.
Hundreds
Of unstructured files processed and contextualized using the agency’s specific logic and preferences.
The Challenge
Each business unit grew up with its own documents, drives, and habits, so the agency’s standards were spread across all of them. How to run a client kickoff, what the brand voice sounds like, and which priorities leadership cares about this quarter lived in decks, spreadsheets, and all-hands recordings, and in the heads of the people who’d been there longest.
The internal site meant to hold it all had fallen behind after the person who maintained it left, and the operations team estimated about half of it was accurate. Everything else circulated over email and Slack. When the agency rolled out a general-purpose AI notebook tool, the operations lead ended up feeding it documents by hand to keep it useful, trading one maintenance job for another. New hires learned by asking whoever was nearby, and teams sometimes worked from guidance that was out of date.
The agency wanted to turn its know-how into a product that could carry its standards to every employee and keep itself current.
The Solution
The agency now runs its own branded product on Lemma, live behind its existing IdP sign-in across every business unit. It carries the agency’s playbook to its people, meaning new hires learn how the agency works from their first day, a process change reaches every team the same way, and teams can check their work against the agency’s brand standards and current priorities. Lemma Workers keep it current behind the scenes, pulling new and updated files from Google Drive as they change, so there’s no publishing step and no one feeding it documents by hand.
Because every insight delivered cites the documents behind it, the agency stays in control of what the product says. When a source file is updated, the guidance changes everywhere. Before any change to the model or prompts goes live, it’s tested against a set of questions approved by subject-matter experts. The agency launched to every employee at once, and in the first seven weeks, a third of them ran nearly 2,600 queries without being told to use it. Agency-wide marketing expertise and best practices can be pointed at every deliverable, providing executive-level feedback and guidance to all levels of employees.
The product also gives leadership an anonymized view of what teams are curious about, something they previously didn’t have. The first monthly themes report surfaced a cluster of questions about a temporary office move, so leadership could address it before it became speculation.
Lemma Capabilities
Their product, their brand, their IP.
The application runs under the agency’s own name, behind its existing sign-in, across every business unit, and the knowledge behind it stays under the agency’s control. Lemma runs as the infrastructure.
Keeps itself current.
Lemma Workers sort decks, spreadsheets, and documents from Google Drive, read each file type on its own terms, and make new content available within about two hours.
Speaks the agency’s language.
Questions are read in the context of the conversation and the agency’s own terms, and responses can draw on several documents at once. Employees don’t need to know where something lives to find it.
Thread AI support throughout the build.
Given the agency’s bandwidth, Thread AI’s Applied AI team supported the product from scoping through launch and stays on hand as it grows.
What’s Next
As the agency adds new processes, brand guidance, and priorities, each one can reach every employee through the same product, so its expertise grows with the business and no longer depends on the people who’ve been there longest. Over time, the product can take on more of what the agency knows, giving every business unit and every employee the same starting point for client work.
Does this sound familiar?
Your know-how is spread across teams, tools, and people, and every group stores it differently.
Someone on your team hand-feeds documents into an AI tool to keep it useful.
You want to build your own AI product but aren’t able to staff an engineering team to build and maintain it.
Your expertise is your edge, and you want to run it like a product, not a shared drive.