Sam Jenkins, Managing Partner at Punchcard Systems, has spent more than two decades working at the intersection of technology, entrepreneurship, and organizational transformation. Since co-founding Punchcard Systems in 2016, he has helped guide the company’s growth while advising organizations on digital workplace strategy, custom software, artificial intelligence, and emerging technologies. His earlier experience includes co-founding Startup Edmonton and Wellnext, leading business development and Microsoft workplace technology initiatives at iomer internet solutions, and serving in executive and board positions across technology, healthcare, public safety, and community organizations. This background has given him a broad perspective on how technology changes the way people collaborate, make decisions, and perform their work.
Punchcard Systems is a Canadian software and technology partner that works with scale-ups and midsized enterprises to design, develop, and implement digital solutions tied to measurable business objectives. Serving more than 150 clients across Canada and the United States, the company’s capabilities include custom web and mobile software, artificial intelligence, cloud infrastructure, data and analytics, user experience design, and workplace technologies such as Microsoft 365, Azure, Copilot, Teams, SharePoint, and Power Platform. Punchcard focuses on helping organizations modernize operations, improve employee and customer experiences, and integrate new technologies into practical business processes.
Your career has followed the evolution of workplace technology, from Microsoft SharePoint-based employee platforms and co-founding Startup Edmonton to building Wellnext and eventually Punchcard Systems. How has that experience shaped your view of the shift from software that employees operate themselves to AI agents that increasingly perform work on their behalf?
When I first started working with digital workplace technologies like SharePoint (back in 2003!), the focus was on creating better ways for employees to find information, share knowledge, and collaborate. At the end of the day, the employee still did the work: they navigated systems, they produced the outputs. Today, we’re moving beyond AI as an individual productivity tool and into a new era of agentic AI, where systems can take on more complex, multi-step tasks on behalf of employees.
That’s the part that’s changing. As an ecosystem, we’re moving past AI as a personal productivity tool into agentic systems that take on complex work on behalf of real humans. Claude Cowork and ChatGPT Work are both making that agentic approach pretty mainstream.
I’ve watched a few of these technology shifts now, and the pattern holds every time. The organizations that win don’t treat new technology simply as a deployment. They treat it as an opportunity to rethink how people, process, and technology work together.
ChatGPT Work can gather information from connected applications and files, complete multi-step assignments, and produce finished documents, spreadsheets, and presentations. How will this change the daily responsibilities of knowledge workers who previously created these materials themselves?
I think that the work is shifting from “producing outputs” to “governing outcomes.” This fundamentally changes the role of knowledge workers, but this doesn’t replace the need for human insight and oversight. Instead, employees will increasingly shift from creating every piece of work themselves to managing, reviewing, and orchestrating AI-generated outputs.
A lot of the repetitive work will move to the agent: gathering information, building presentations or documents, or summarizing data. This frees people up for the work that actually needs a person: strategy, creativity, critical thinking, and collaboration.
The skill that becomes increasingly valuable is knowing how to direct AI well, by giving it the right context, and recognizing when the answer looks right but isn’t.
You have described a growing need for “AI managers.” What would this role involve in practice, and how would it differ from the responsibilities of an IT administrator, product manager, or traditional people manager?
AI managers will play an important role in helping organizations adopt AI responsibly and effectively. Their responsibility will be to manage and govern AI outputs, ensuring that AI-generated work is accurate, reliable, and aligned with business objectives.
This is a different job from traditional IT, which owns infrastructure and technical systems, or a product manager, who focuses on developing and improving products. Every department is going to need someone who understands how AI applies to their specific ares, whether that’s finance, legal, HR, or operations.
The real shift is ownership. Right now, most organizations haven’t designated someone who is clearly accountable for whether these systems deliver value and stay inside acceptable risk. That gap, in terms of identifying directly responsible individuals, is what these types of roles will fill.
KPMG Canada found that AI agents are already influencing how organizations hire both entry-level and experienced employees. Which skills will become more valuable as employers begin assessing candidates on their ability to delegate work to AI, evaluate its output, and intervene when necessary?
The skills that will become increasingly valuable are the ones that remain uniquely human, such as critical thinking, judgement, communication, creativity, problem-solving, and leadership.
Technical AI literacy will continue to matter, but employers are looking for people who understand how AI fits into the broader business environment. That includes skills like change management, risk management, and strategic thinking.
The line between technical and non-technical roles is blurring. Organizations are moving away from hiring only for the skills someone already has and toward people who are willing to learn, adapt, and continuously upskill. The most valuable employees will be those who can use AI capabilities while understanding where human judgement remains essential.
Many employees traditionally developed professional judgment by completing repetitive entry-level tasks. If AI agents absorb much of that foundational work, how can organizations redesign training and career development so younger employees still gain the experience needed to become effective decision-makers?
This is a truth that worries me a bit. People used to build judgement and experience by grinding through entry-level tasks. If agents are doing that work, the ladder loses a few rungs.
Organizations will have to be more deliberate than ever about building junior talent through mentorship, coaching, decision-making exercises, and exposure to higher-value projects earlier in someone’s career.
There is also an opportunity for reverse mentorship, where younger employees who are more familiar with AI tools can help teams understand how these technologies can be applied, while experienced employees continue to provide the business context and judgment that AI cannot replicate.
The goal should be to redesign the learning process for a workplace where humans and AI work together.
Punchcard has extensive experience implementing Microsoft 365 and related workplace technologies. How does ChatGPT Work change the competitive landscape for Microsoft Copilot and Google Gemini, and what factors should organizations consider when selecting an AI platform for everyday work?
The landscape is evolving quickly, and organizations have more options than ever. The decision should come down to which solution best aligns with an organization’s existing technology environment, workflows, security requirements, and business goals.
For organizations already invested in Microsoft 365, tools like Copilot may provide advantages through integration with existing applications and data. Others may find different platforms better suited to their needs.
I think the key question that leaders should be asking isn’t “which tool?” It’s “what problem are we solving, and which platform helps us solve it responsibly?”
Here’s what we’ve learned first-hand: successful adoption is about more than the tool. We’ve implemented company-wide AI platforms, with custom models tuned to how we actually work. Every employee uses it, but that didn’t come from picking the right product; it came from strong data foundations, governance, and building AI into how the company runs.
Model performance often dominates conversations about workplace AI, but organizations also need to consider permissions, data access, auditability, and accountability. What governance controls should be established before an AI agent is allowed to operate across email, documents, internal databases, and business applications?
One of the biggest mistakes I see organizations make is treating AI governance as an afterthought. As AI systems become more capable and connected to business data, organizations need to establish clear principles around how these tools are used before scaling them broadly.
That starts with understanding what information AI systems can access, ensuring permissions are appropriately scoped, and putting controls in place around monitoring and accountability.
Organizations also need clear guidelines around where AI can operate independently and where human oversight is required.
By building the right foundations around data and security upfront, businesses can unlock the value of AI while managing the risks that come with increasingly autonomous systems.
As AI agents take actions rather than simply provide answers, mistakes can have operational consequences. How should businesses validate agent-generated work, determine which actions require human approval, and assign responsibility when an agent produces an inaccurate result or takes an unintended action?
When AI moves from answering to acting, the risk profile changes. Not every task carries the same stakes, so the bar for human sign-off should scale with what’s riding on the outcome.
Simply put, this requires clear ownership. AI should support decision-making, not replace responsibility. Organizations need to define who monitors AI performance, validates outputs, and is ultimately accountable for results. The goal is to combine AI-driven efficiency with human judgment, creating systems that are both scalable and trustworthy.
Many organizations are experimenting with AI without yet connecting adoption to measurable business outcomes. How should leaders identify worthwhile agentic use cases, measure their return on investment, and determine whether a pilot is ready to scale across the organization?
The gap usually traces back to sequencing. Teams pick a technology first and then hunt for somewhere to put it, when the durable wins come from starting with a workflow that is already painful and expensive and asking whether an agent is genuinely the right tool.
The best filter is variability. If a process is predictable enough for a rules engine, you don’t need an agent. If it’s so open-ended that no one can describe a good outcome, the agent will struggle and you won’t be able to trust it. The value sits in between, in work that needs judgment across several systems but still runs inside bounded, auditable rules.
I’d add one screening question that changes the risk profile more than anything else. Is the cost of a wrong action bounded and reversible? Where it is, let the agent act and move fast. Where a mistake is expensive or hard to undo, keep a person in the decision.
On return, the common mistake is measuring the wrong thing too late. Hours saved is the default metric, but it undersells the impact. The larger returns show up as cycle-time compression, lower error rates, and capacity that was previously rationed becoming available. A support team that could only reach half its tickets becomes a different business once it can reach all of them. Capture a clean baseline before the pilot starts, because without it you’re stuck arguing over anecdotes. And when you tally cost, count the full picture. Integration work and human oversight time are real line items, and oversight is chronically underestimated.
Deciding whether to scale is where discipline matters most, and it’s the step teams rush. A demo on clean inputs tells you little. A pilot is ready when it holds up on the long tail of messy, real inputs, when the failure modes are understood and bounded, when observability is in place so a silent failure gets caught rather than discovered by a customer, and when a named person is accountable in production.
Run a unit-economics check too, since agent costs scale with usage in a way software seats do not.
Agentic systems fail differently than traditional software. They fail probabilistically and often quietly, so the organizations that pull ahead build evaluation and monitoring in from day one rather than bolting it on after something breaks.
Punchcard Systems helps organizations move from experimenting with AI to integrating it into real business processes. How do you assess an organization’s readiness for AI, identify the workflows where it can deliver measurable value, and help teams implement these systems responsibly without disrupting existing operations?
It’s important to start by understanding an organization’s current state. One of the biggest misconceptions is that AI adoption is simply about choosing the right tool. In reality, successful implementation depends on having strong foundations: quality data, clear processes, and defined ownership.
From there, we help organizations pinpoint where AI can create the most value, often in workflows that eat up disproportionate time relative to their impact, or in decisions being made without the right information on hand.
Organizations have to think about how roles and decision-making will evolve alongside the technology. By combining the right technology with employee adoption and change management, businesses can integrate AI in a way that creates measurable value without breaking what already works.
Thank you for the great interview, readers who wish to learn more should visit Punchcard Systems.
