There’s a pattern I’ve been watching for the past two years across the Indian SME landscape, and it’s worth naming out loud.
A founder attends a conference, comes back convinced the company needs to “do something with AI,” and tasks the HR team with figuring it out. They trial a chatbot. They subscribe to an AI-enabled HRMS module that they don’t have the bandwidth to configure. Six months later, the tool is abandoned or running at 10% capacity. And the conclusion? “AI isn’t ready for HR.”
That conclusion is wrong. The implementation was.
The Real Problem: The HRMS Trap
Most companies with 50 to 300 employees already have an HRMS in place. These systems are capable. They come with performance management modules, engagement surveys, and AI-assisted analytics. On paper, they should be solving the problem.
In practice, most organizations are using 20–30% of what they’ve paid for.
HRMS platforms are built for standardized workflows. The moment an organization needs something that doesn’t fit the default – a custom appraisal rubric, a performance framework tied to specific business metrics – the system either can’t accommodate it or requires expensive customisation. Hidden costs, including module-gating, data migration, and customisation fees, can quietly add 30–60% to the quoted price.
I’ve worked with companies that paid for a performance management module, then built a parallel Excel process because the system’s output didn’t match what leadership actually needed. The HRMS licence is still being paid. The problem is still unsolved.
This is the core issue: the savings AI promises – in admin time, decision consistency, process speed – don’t materialise when the underlying workflow is locked inside a system that isn’t being used fully. You pay for two overlapping systems and get the full benefit of neither.
Where AI Actually Delivers
The organisations that extract genuine value from AI in HR apply it to defined, documented workflows – not the other way around.
Three areas where the return is clear and immediate:
Structured performance decisions. Reviews in most SMEs are inconsistent across managers – not because managers are poor at their jobs, but because there’s no scaffold for consistency. AI-assisted frameworks where managers input observations, and the system structures feedback across defined parameters – productivity, customer impact, team contribution – produce complete, comparable, documented reviews. Quality improves. Time per review drops. HR gets data it can actually use.
People analytics on data you already have. Attrition signals, engagement dips, workload imbalances – these patterns exist in data most SMEs already have in their HRMS or attendance systems. The gap is not the data; it is the bandwidth to analyse it. AI applied to leave patterns, and performance trends can surface early warning signals three to four months before they would otherwise surface. Most Indian SMEs have never used this capability – not because it doesn’t exist in their system, but because nobody configured it.
Intelligent content generation. Policy documents, offer letters, performance communication, exit letters – these consume disproportionate HR bandwidth in lean teams. Generative AI configured with your compliance requirements, tone, and document standards produces first drafts that are 80% ready. For a two-person HR team managing 150 employees, this leverage is immediate.
The Uncomfortable Truth
Most Indian SMEs don’t have the process foundations for AI to work on.
No documented job descriptions. Performance criteria that live in the MD’s head. Compensation decided informally. Exit processes that vary based on who’s handling it that week.
AI amplifies what exists – clean process or broken process alike. If the inputs are undocumented and inconsistent, the output is unreliable regardless of how sophisticated the model is.
This doesn’t mean fixing everything before you start. It means being honest about where your processes are defined well enough for AI to add value – and beginning there. One high-frequency, high-effort workflow. Document it. Apply AI to it. Measure before and after. Then expand.
The organisations that fail at AI in HR almost always skipped this step.
The Opportunity Is Real – But It Requires Discipline
India’s MSME sector employs over 230 million people. The vast majority of these businesses have one or two HR professionals managing the people function for hundreds of employees. That ratio is unsustainable without leverage.
AI is that leverage – but only when applied to something solid.
The businesses that will pull ahead are not the ones that bought the most sophisticated HR technology. They are the ones that built clean enough processes for AI to work on, then used it to make those processes faster, more consistent, and more intelligent.
The right question isn’t “which AI tool should we buy?”
It’s “do we have something worth automating?”
Start there.