Beyond AI Adoption: Sameer Nigam, CEO & Co-Founder of Stratbeans, on Turning AI Investments into Enterprise Value
The next phase of enterprise AI transformation will not be defined by how quickly organisations adopt technology, but by how effectively they translate that investment into measurable business outcomes. As AI becomes increasingly embedded in enterprise operations, the real opportunity lies in aligning workforce capabilities, intelligent workflows, and leadership priorities to unlock sustainable productivity and performance.
Strengthening its thought leadership in this evolving landscape, brings a forward-looking perspective to CXO Today, where Sameer Nigam, CEO & Co-Founder of Stratbeans, shares his insights on overcoming the AI productivity paradox and turning technological potential into tangible enterprise value.
1. As AI adoption moves from pilots to enterprise-wide deployment, what capability-building does the organisation actually need to make that transition pay off?
Sameer: Most companies will discover that adoption is not the bottleneck — capture is.
A 2026 NBER study of nearly 6,000 senior executives across four countries found 89% reported no measurable impact from AI on productivity over the preceding three years. These were executive estimates of sales per employee, so this is not proof that AI has no value. It is evidence that the value is getting stuck somewhere between the tool and the P&L and that location is almost always the same one economists identified during the earlier IT productivity paradox: technology gains require complementary changes in process, authority and incentive design before they surface as output.
An employee may draft a proposal faster, but revenue will not arrive sooner if pricing approval still takes a week. Workforce development has to travel with changes to workflow and decision rights, not ahead of them or behind them.
I would start with a specific constraint: slow onboarding, excessive escalations and work backward to the capability that removes it. Employees need practice in real situations; managers need to learn how to run the redesigned work, not just approve it. The test for scaling is whether the business can deliver more valuable output, at the required quality, after accounting for the full cost of change. That gives learning investment to a commercial owner, not just a budget line.
2. How should enterprises identify the skill gaps that actually matter, and personalise learning against them?
Sameer: Not every skill gap is a business problem. A small improvement in a capability that constrains output is worth more than a large improvement in one that does not.
Take a sales team with strong product knowledge but weak discovery skills. More product training will raise assessment scores and do nothing for conversion. AI is useful here for a specific reason: it can analyse approved interaction samples and live assessments to isolate the exact behaviour that needs attention, with managers validating the diagnosis before any content gets built.
What has changed recently is how fast that diagnosis can turn into a fix. Closing a skill gap used to mean routing it through a content development team a brief, a queue, a production cycle measured in weeks. Articulate, the content-authoring platform we represent exclusively in India, launched a product called Frontline at Articuland Orlando this September, built specifically to remove that queue. A sales head, a support head, or a corporate communications head can now turn their own material directly into structured training, without routing through a content specialist. The diagnosis and the fix move at the same speed for the first time.
That is the shift worth watching, not because the tool itself is remarkable, but because of what it reveals: the real bottleneck in closing a skill gap was rarely the diagnosis. It was production capacity. Remove that constraint and the question becomes whether the organisation has the discipline to decide what is worth building quickly, rather than simply the ability to build it.
For a CEO, the metric that matters is time to independent productivity, and how much expert supervision a new hire requires along the way. If a company hires 200 people a year and cuts time to proficiency by one month, it has brought forward 200 employee-months of productive capacity, the financial value of which depends entirely on what those employees can produce during that window.
3. How can learning-in-the-flow-of-work build new capability without slowing people down?
Sameer: Its case rests on a single number: how much productive time disappears into search, waiting, and avoidable error and whether assistance at the point of need gets that time back.
The 2025 study Generative AI at Work (Brynjolfsson, Li and Raymond), covering 5,172 customer-support agents, found a 15% average increase in issues resolved per hour. That is not a training gap closing it is a retrieval gap closing. The employee generally knew the answer existed somewhere; the constraint was getting to it fast enough to use it.
The more interesting question is what the business does with the capacity that frees up. As an illustration: 500 employees saving 30 minutes a day across 220 working days release 55,000 hours a year. That capacity only acquires financial value if it serves additional demand, reduces overtime, or avoids a hire the business would otherwise have made.
One design discipline matters here: good assistance should show its reasoning, not just its answer. Otherwise employees get faster at following instructions without getting better at handling the unfamiliar. I would track both — immediate performance, and whether independent capability is actually growing underneath it.
4. What is the real obstacle to adoption of AI-enabled learning, and what should leaders do about engagement?
Sameer: Most adoption failures are not training failures. They are incentive failures.
If employees believe that surfacing a productivity gain will simply raise their target or put their role at risk, they have a rational reason to sit on useful discoveries. Leaders need to be explicit about how productivity gains will be redeployed and what development path follows and that conversation only works if it is backed by real opportunities, not a slide.
Manager incentives matter just as much. A manager measured purely on this week’s output has no reason to release people for practice whose payoff lands next quarter. The business plan has to price in the upfront cost of practice against the return that follows, or the manager’s rational choice will always be to withhold people from learning.
I would start with a problem employees already want solved, design the intervention with them, and make the benefit visible quickly — fewer escalations, faster resolution on hard cases. Attendance and login data tell you almost nothing. Repeat use and demonstrated capability tell you everything.
5. How does AI help enterprises capture institutional knowledge and make expertise accessible across the business?
Sameer: Institutional knowledge sets the speed limit on growth. If every new branch, product, or complex customer case still requires the same three experts, growth is really just pressure accumulating on three people.
The knowledge worth capturing is rarely the procedure — it is the judgment behind the exception: why this request needed escalation, what signal flagged a problem early, when the standard process stopped being the right one. That reasoning is what makes expertise reusable; a document library of procedures is not.
AI is useful for retrieving the relevant guidance and turning validated cases into practice material provided employees can see the source, understand its limits, and still reach an accountable expert when the case warrants it.
One real risk follows directly from how fast content creation itself is decentralising. When a sales head or a support head can produce a training kit in an afternoon, as tools like Frontline now allow, governance can fall behind production speed. Interestingly, Articulate has built this concern into the product itself, keeping a layer of brand standards, review and analytics sitting above the content so L&D still owns quality even as creation moves outward. That is the right instinct generally: decentralising creation without strengthening governance in parallel is how institutional knowledge quietly curdles into institutional noise. Ownership and regular review are non-negotiable regardless of who is doing the creating.
At Stratbeans, I treat learning and knowledge management as two views of the same constraint: repeated questions show you where people need development; well-resolved cases are the raw material that teaches everyone else. The real test is whether the business can grow while reducing the expert time required per transaction, employee, or location.
6. As AI reshapes roles, what should CXOs prioritise to build both AI-ready employees and durable human capability — judgment, collaboration, critical thinking?
Sameer: Judgment, above everything, because it decides whether speed creates value or creates expensive mistakes at scale.
The Harvard-BCG “jagged frontier” study makes the stakes concrete. In a field experiment with 758 consultants, those using AI completed suitable tasks roughly 25% faster. On a task deliberately placed outside the system’s effective capability, the same consultants were 19 percentage points less likely to reach the correct answer than those working without AI at all. The sample is management consultants, not every function, so I would not assume the exact numbers travel unchanged to a factory floor or a back office but the underlying pattern, that confidence and correctness diverge at the edges of AI’s competence, shows up everywhere we have looked.
Learning programmes should put people in front of realistic decisions with incomplete information and real consequences, and require them to defend their reasoning, challenge a plausible but wrong recommendation, and reach a conclusion with others, not just evaluate a finished answer.
There is a longer-run workforce question underneath this. Junior employees traditionally built judgment through research, first drafts, early analysis. As AI absorbs more of that work, leaders have to deliberately preserve the practice reps that used to happen by default. Otherwise a company improves this year’s efficiency while quietly running down the supply of judgment it will need five years from now. A workforce strategy that only optimises today’s output is not a strategy, it is a withdrawal from an account nobody is refilling.