How our account team is building better context for AI

Sixty-six percent of knowledge workers say AI gives them more time for high-value work. Eighty-five percent of AI users say it helps them focus on what matters most. And 82% say they still lose too much time just searching for information they already have somewhere.

That last insight matters to our account team. AI can save time on a task. It can’t save the time spent searching for a file, decision or approval that already exists somewhere in your systems.

This is the fifth piece in our series on where AI actually stands. The first four covered what clients want, our company-wide approach and how our copy and creative teams rebuilt their processes. This one is about our account team and a harder challenge they uncovered: making AI genuinely useful when it doesn’t have the current, relevant or appropriate client context.

Where the account team started

Our account team started where most teams do: using LLMs and everyday embedded AI tools to speed up meeting notes, transcript synthesis, research, follow-up emails and calendar searches. From there, the team moved into testing prompts for account-specific work: client intake, structured briefs, timelines and early support for drafting statements of work.

The real bottleneck

AI can support a task, but it doesn’t automatically know the client, the industry, the audiences, the project history or the outcomes that have already worked. Without that context, every query still starts from zero, requiring repetitive manual entry before the tool can produce useful support. That’s exactly the kind of work the team was trying to reduce in the first place.

Knowledge is hard to access for a more basic reason: Even in highly organized, well-governed teams, past work, decisions, feedback and account knowledge can still end up scattered across files, platforms and stakeholders. It happens naturally as projects move quickly, team members develop different working styles and people transition in and out of accounts. Better context leads to better support. Organizing client information in one shared place helps the team use AI more consistently and specifically, while preserving institutional knowledge beyond any one person’s memory. That’s what pushed us to look for a different kind of solution than simply adding another tool.

What we’re building

The solution wasn’t another disconnected AI experiment. It was a better context system. We’re building a curated client knowledge hub that brings important account context together in one secure, governed place. That includes client background and key stakeholders, brand guidelines and approved messaging, past deliverables and recommendations, client feedback and approved decisions, statements of work, estimates and project history. All this information lives within the approved tools our teams already use, supporting responsible AI use without moving client knowledge into unapproved environments.

The goal isn’t always to adopt new tools. It’s about developing smart systems and structures to organize a shared source of truth, so the team can work from the right account context instead of rebuilding a foundational brand knowledge base with every project. It also helps new team members get up to speed rapidly without waiting on someone else’s memory.

What this makes possible

Less duplicate work. Faster onboarding for new team members. More consistency across accounts. Most importantly, institutional knowledge stays with the account even as team members move into new roles, helping preserve continuity and context over time within the approved systems and safeguards our teams already use.

What’s ahead

Notion Partners is expanding this approach across client accounts, using secure, approved systems to help new projects get off the ground faster, sharpen creative briefs from the start and eventually support smoother handoffs between teams. The bigger opportunity is not just using AI to move faster. It is giving teams better access to the right context, so the work starts smarter, stays more consistent and carries forward with less rework.

This is the fifth piece in our series on where AI actually stands. Next and last, we hand it over to our video team to explore how they built a way to stop settling for stock imagery entirely.

This article was originally published on LinkedIn on September 16 , 2026.