“Garbage in, garbage out” has long been a reality in enterprise data management. But with the rise of autonomous AI agents, the consequences of poor data have never been greater.
Unlike traditional analytics or dashboards, agents don’t just interpret information; they act on it. And if their decisions are based on fragmented, inconsistent, or siloed data, enterprises risk accelerating inefficiency rather than eliminating it.
In a recent fireside conversation, Planview CEO Razat Gaurav sat down with Alan Manuel, GVP of Product Management, to explore how enterprises can prepare their data foundations for the era of agentic AI. You can watch their full discussion below:
Alan noted throughout the conversation that while individuals may be working at capacity, customers often don’t see outcomes because the system isn’t aligned. That’s why, he argued, platforms require more than raw access to information; they need structured, governed, and semantically consistent data to produce trustworthy results.
Why “garbage in, garbage out” is magnified with autonomous AI
Planview’s research into more than 3,500 value streams shows just how vast the efficiency gap can be. According to Alan, “For every dollar a typical company puts into software development, only about 25 cents actually translates into business results. The rest is lost to misalignment, teams working on the wrong things, or waste in one form or another.”
Autonomous agents can make this worse if they rely on incomplete or siloed data. An agent might plan more work than teams can realistically deliver, ignore bottlenecks between groups, or chase outcomes that are disconnected from strategic goals.
Without a unified view of capacity and value, AI risks reinforcing the very problems it promises to solve. A unified view of capacity and value means knowing who can do the work, and whether it’s the right work to begin with. Without that connection, teams stay busy, but customers see little change. When you line up resources with real outcomes, every hour spent, whether by a developer or an AI agent, moves the business forward.
Three Foundational Layers for Enterprise-Ready AI Agents
To build AI agents leaders can trust, enterprises need (in part) a governed, connected data architecture. Alan outlined three critical layers:
Data foundation
Agents need visibility into all enterprise systems, not just isolated silos. “If you’re planning for more than your actual productive capacity, which 99% of companies not using Planview are doing, you’re setting yourself up for failure,” Alan explained. Editorially, this is where Planview’s Flow Framework, with 60+ pre-built connectors, comes in — providing the breadth of inputs needed to understand investments, capacity, and work in flight.
Governance layer
Autonomy without control is dangerous. A strong governance layer enforces permissions, security, and context so agents act within safe boundaries. Product managers, for example, can let AI recommend backlog priorities without exposing sensitive financials.
Alignment Layer
Beyond data breadth and governance, enterprises need a way to connect strategy to execution. Alan emphasized that OKRs act like the “nervous system” of the organization, carrying signals from strategic portfolios down to individual teams. This alignment ensures each group understands what part of the problem they’re solving, preventing strategy from stalling at the top and keeping daily work connected to business outcomes.
Codifying business logic through alignment
This alignment layer also codifies business logic — the rules of the road for AI reasoning. For example, if a company sets a strategic goal to improve customer experience, OKRs help translate that into measurable outcomes for each team. With strategic portfolio priorities, OKRs, and team-level planning frameworks in place, an agent can recommend backlog adjustments or highlight risks in meeting objectives. Without them, strategy is easily lost in translation between executives and developers.
From outputs to outcomes
Another recurring theme in the discussion was the shift from outputs to outcomes. Developers don’t want to work endlessly on low-value backlog items; CFOs don’t want activity for its own sake. Both want meaningful progress toward strategic goals.
Alan described OKRs as the “nervous system” of the enterprise — carrying signals from the strategic portfolio (the brain) to every team in the organization. “Each team needs to know what part of the problem they’re solving,” he said. “Otherwise, strategy isn’t worth the electrons it’s stored in.”
Agentic AI becomes powerful when it can connect these layers: strategy portfolios, cascading OKRs, and team-level planning. With trustworthy data, agents can surface bottlenecks, rebalance work in line with strategic goals, and keep the whole system moving toward outcomes rather than outputs.
The opportunity ahead
As Razat framed in the conversation, technology executives are under pressure to deliver more productivity, more velocity, and better quality, all while justifying rising levels of investment. Autonomous agents can help close the gap, but only if they stand on a solid data architecture.
And, as Alan summed up from a different angle: “If you want to make developers happy, and developers are a scarce resource, make sure they’re working on meaningful, high-value items. That’s what the CFO wants, and that’s what the developers want.”
Agentic AI is not a shortcut. It is an amplifier. Enterprises that invest in governed, connected, 360-degree data today will be the ones who unlock its full potential tomorrow.
See how you can drive faster, smarter decision-making with Planview AnviTM, your next-generation AI for connected work. Learn more
