Failing to provide context stalls ROI and invites risk, but there’s a way to get the best out of your agents.
Sydney Scott
Editorial Strategist, AI
Workday
Without situational awareness, a sophisticated agent is simply an expensive dead end.
Research from Carnegie Mellon University highlights that while agents succeed at nearly 60% of simple tasks, their success rate plummets to just 35% when faced with multi-step workflows. There could be any reason for this failure, but quite often it’s a lack of context.
For an agent to be truly effective, it must perceive its environment and learn from past interactions through a digital backstory or context that allows it to remember previous events and specific business rules.
Context is the essential foundation for intelligence—the background that keeps an AI agent from drifting into irrelevance. It is a digital framework of history, business rules, and specific roles that allows an agent to act reliably within your company. Without this foundation, you’re accumulating rework and creating unreliable systems that eventually generate more work than they automate.
Nearly 40% of the productivity promised by AI is being silently erased by rework.
Report
The secret to a great AI agent isn't just speed, it’s a good memory. When we give agents the right background information, they can save us hours of time. Without this history, research from Workday shows that we lose 40% of our productivity because we have to spend time fixing the AI's mistakes. For every 10 hours the AI saves us, we spend four hours correcting it.
Giving your AI a backstory keeps it on track and makes it a reliable part of the team. This helps your business in three big ways:
Doing the job right the first time: Agents with a memory won't repeat tasks they already finished or treat a loyal, long-term customer like a total stranger.
Saving money and mistakes: It stops the AI from getting stuck in loops that waste expensive computer power. It also prevents the AI from signing bad contracts or breaking company rules.
Keeping things safe: An agent that understands its role is less likely to accidentally share private passwords or get tricked by hackers.
When AI understands the why and how of your business, it stops being a confusing tool and becomes a useful partner. This is how you avoid becoming one of the 95% of AI projects MIT researchers say fail to reach their goals. By focusing on context, orgs can ensure their agents reduce the workload instead of creating needless rework.
Up to 95% of corporate AI projects are failing to deliver their intended value.
To avoid these risks and build agents worthy of trust, enterprises must move from flashy demos to structured context engineering. This framework relies on three foundational pillars that turn a powerful model into a reliable partner.
Defining an agent's profile and persona—its role, responsibilities, and ethical boundaries—is a functional constraint that governs behavior. By establishing a specific backstory, such as telling your agent they are a financial analyst with a decade of expertise, the system can tailor its depth and tone to match professional expectations. This consistency is a direct revenue driver. For instance, McKinsey notes that organizations that use AI to create tailored experiences for customers can enhance satisfaction by 20%. It’s the difference between sending a generic email versus a carefully curated list of recommendations. Companies that give their agents the necessary backstory for these interactions can see sales ROI increases of up to 20%.
For an AI agent to use context, it first has to find it. Standard search tools often fail because they only look for keywords, not the why. Leading companies use knowledge graphs—a map of how facts connect—and context graphs—a map of how decisions are made. This organized memory prevents context rot and ensures the AI acts on facts, not guesswork. It turns company knowledge into a digital memory, protecting them from losing vital information.
For hard business projects, one AI agent can get overwhelmed and start making mistakes. The best solution is to split the work among a team of specialized agents that talk to each other to plan and solve problems. Orchestration is like a traffic controller; it makes sure each agent gets only the specific information it needs to do its part without getting confused. This teamwork also saves money—companies can use more powerful AI for the hardest thinking while using smaller, faster AI for simple tasks to speed things up.
By building a complete backstory, companies can avoid creating messy, expensive systems that don't work.
Leaders must stop viewing AI agents as mere task-replacement tools. Treating agents as integrated members of the workforce isn't a marketing metaphor—it is a strategic requirement for efficacy. To unlock their full value, we must treat them as true collaborators, providing the same rigorous training and clear backstory required for any high-performing hire.
Building a strong digital workforce isn't about trying to fix everything at once. It’s about taking smart steps: starting with clean data and moving toward teams of specialized agents that handle entire business processes. By building a complete backstory, companies can avoid creating messy, expensive systems that don't work. Instead, they can build AI that fits perfectly into how the business actually runs.
The impact of getting this right is huge. Research from MIT Sloan experts suggests that this year marks a critical shift for leaders—moving from simple experimentation to building AI solutions that create real value and competitive advantage at scale. Giving agents a memory of how your company works can do that.
Now is the moment to move beyond experimental pilots. Prioritize your context strategy today to pioneer a future where AI and human insight work in perfect harmony to drive the next era of innovation.
Report
AI has been a gamechanger across nearly every industry and marketplace. But realizing the true value of AI requires buy-in between different stakeholders. In this AI Masterclass, learn how you can make the most of AI for finance, HR, and more.