Rebuilding packaging workflows with AI

Sriram has spent 25 years leading digital transformation, data, and AI projects for Fortune 500 companies, and he frames the current moment as bigger than any technology cycle he's worked through.

AI, he says, is starting to reshape how organizations operate, compete, and create value, and the distance between the companies acting on it and the ones waiting is expanding daily.

His advice to teams sitting on the fence is blunt: don't wait for the perfect moment or perfect clarity because starting, in packaging specifically, even small improvements across the lifecycle compound into a large impact at scale, which is why the cost of waiting is higher than it looks.

The biggest near-term shift, according to Sriram, is the removal of friction that artwork operators, project managers, compliance managers, and QC analysts deal with every day. A large share of their time goes to work they shouldn't have to do, such as reorganizing briefs that arrive in inconsistent formats, hunting for and cleaning files, or bridging systems that were never built to talk to each other.

His framing is that AI is an opportunity rather than a threat: it strips out that grunt work and gives people their time back for strategic thinking. He returns to a line he likes as the clearest way to settle the fear about job loss.

Sriram points to two concrete bottlenecks. The first is briefing: client briefs arrive as Word documents, templates, emails, and other unstructured formats, and the nuance gets lost or re-interpreted person by person. AI can ingest those briefs, structure them, and pass clear instructions downstream so everyone is working from the same understanding.

The second is quality control. When a regulation changes, hundreds of SKUs may need updating, and each one has to be checked. Today, that is done largely by hand. That manual QC is the biggest bottleneck in moving a brand from idea to shelf, which is why 5FLOW is building AI-assisted QC to speed it up.

He's also clear that point fixes aren't enough. If one team adopts AI and produces content five times faster while the next team still reviews everything manually, the end-to-end process hasn't improved: the bottleneck has simply moved from step one to step two. The real gain comes from reimagining the workflow as AI-native, end to end.

Adoption starts slow, Sriram says, but it follows a hockey-stick curve once AI is embedded in the workflow and people see bottlenecks disappear. He lays out a four-step path: first, give people AI as a standalone tool to get comfortable and remove the fear; second, integrate AI into the workflow so it's seamless rather than optional; third, move to AI-led but human-governed operations, where the real value at scale appears; and fourth, do it faster still, freeing people almost entirely for strategic work. Most organizations, he estimates, are somewhere between steps one and two.

He grounds this in a recent example from Propelis. After rolling out Copilot with 1,000 enterprise licenses to employees worldwide, teams built 100 agents within four weeks, saving close to 6,000 hours and delivering an estimated half a million dollars in impact, all before any formal methodology was in place. In his view, adoption itself isn't the obstacle: the only real barrier is the initial inertia of trying something new.

Most companies are still early, Sriram says. He describes a landscape of ad hoc AI usage, missing governance, no shared workflow, and fragmented adoption, with many employees deploying tools on their own while their organizations lack a mandate, security oversight, or any way to measure impact. That governance gap is the piece most often missing.

His message to anyone still unsure how seriously to take this is that the question has already been answered. AI is changing the industry whether or not a given company participates; the only real decision left is whether you actively shape how it affects your business.

5FLOW's product thinking runs on four principles, which Sriram sums up as augment, automate, accelerate, and differentiate: Are we augmenting people with new capabilities? Automating a repetitive process? Accelerating execution? Or differentiating with proprietary solutions in the market? The team applies that lens to every bottleneck across the packaging workflow lifecycle.

He also makes the case for agentic AI as the unlock for scale. Where individual AI tools used to each do a single job in isolation, agentic development connects multiple tools and systems into one ecosystem, letting companies leapfrog and set up for scale. Pilots alone, he argues, don't deliver value; the value shows up when you scale. His closing view is that the future belongs not to the companies experimenting, but to the ones that figure it out and scale.