The CTO’s AI Playbook
From the Desk of Jenny Peng, CTO, Aptean
Table of Contents
FOREWORD
Leading AI Adoption as a CTO pg. 03
PLAY ONE
Disrupt Yourself First pg. 04
PLAY TWO Pilot Relentlessly with Intent pg. 06
PLAY THREE Make Speed Your Strategy pg. 08
PLAY FOUR Own the Intelligence Layer pg. 10
PLAY FIVE Scale Transformation Through Small Changes pg. 12
FINAL THOUGHTS Where We Go from Here pg. 14
FOREWORD
Leading AI Adoption as a CTO
For the first time in my career, I don't know what the development landscape will look like in six months’ time. There seems to be a new artificial intelligence (AI) platform, another tool, and a different approach every week.
For someone who has spent decades building expertise in software development, it’s unsettling. But this discomfort is exactly where we need to begin.
AI adoption requires letting go of previous expectations. The blueprint is being written as we move, and the best way to contribute is to start with intention, discipline, and a willingness to be wrong sometimes.
At Aptean, we've already run hundreds of pilots and retired plenty of ideas that didn't pan out. We’ve built agents, tested intelligence layers, and learned what works by watching our teams use new technology day-in, day-out.
Today, I’d like to share the five “plays” we've used to guide AI adoption at Aptean. They're drawn from what we've tested, refined, and proven inside our own operations. Your context will be different, but the principles hold.
If you're a CTO or CIO who's ready to move from evaluating AI to embedding it across your business, this playbook is for you.
Jenny Peng CTO, Aptean
Disrupt Yourself First
Break Your Own Model Before the Market Does it for You
The instinct when new technology emerges is to ask: How can this complement what we’re already doing? That’s a reasonable question, but it’s also a trap. AI doesn’t just augment existing processes; it replaces them. If you approach adoption as an add-on, you’ll miss the deeper transformation.
If you wait for your processes to become outdated before you challenge them, someone else—a competitor, a new entrant, a faster-moving peer—will do it for you. By then, you’re reacting instead of leading.
The uncomfortable truth is that AI requires you to abandon practices you’ve relied on for years, sometimes decades. For me, that meant accepting that writing code from scratch is no longer the default starting point for development work. For your teams, it might mean rethinking how they structure workflows, make decisions, or prioritize workload.
Make Your Organization “Customer Zero”
At Aptean, we decided early on that we wouldn’t ask customers to adopt AI capabilities we hadn’t tested on ourselves. So we became our own Customer Zero.
Before we launched AppCentral, our AI platform purpose-built for industries, we deployed it internally. Our finance team built agents to automate AR/AP. Our operations group used AI to extract insights from previously siloed data. Our customer support teams embedded intelligence into their daily workflows.
It wasn’t a clean success story where everything worked on the first try. We ran pilots that failed, but each experiment taught us something. By the time we delivered capabilities to customers, we’d already lived with them. That credibility—the ability to say “we run our business on this”— has become one of our most valuable assets.
Create a Feedback Loop
When your teams use AI daily, they become your best source of operational intelligence. They explore edge cases, identify integration challenges, and discover applications you wouldn’t have anticipated. At Aptean, AI is now used in all our meetings, planning, and decisionmaking, which has directly contributed to our product-grade readiness.
Feedback loops help you learn fast, whether you’re building AI capabilities for customers or using them purely for internal operations. You see where friction occurs and where value compounds, and those insights enable you to design and refine intelligent solutions.
Key Plays:
» Challenge your own processes first. Don’t wait for external pressure to force internal change.
» Become Customer Zero. Test AI capabilities on your own operations before deploying them externally.
» Use internal teams as product intelligence. Daily use exposes the gaps that demos never reveal.
» Build credibility through experience. Whether you’re selling AI or using it internally, proven outcomes create trust with your customers, stakeholders, and your own teams.
“AI requires you to abandon practices you’ve relied on for years, sometimes decades.”
PLAY TWO
Pilot Relentlessly
(With Clear Intent)
Treat Experimentation as an Operating Discipline
Most companies approach AI pilots as proof-of-concept exercises: low-stakes experiments designed to evaluate whether something might work. But when you frame pilots as evaluations, teams optimize for safety instead of learning. They avoid ambitious use cases and hedge their commitments.
At Aptean, we’ve run hundreds of AI pilots. Some delivered immediate value and scaled quickly. Others taught us what doesn’t work, or which use cases aren’t ready yet. We learned from both.
Either way, we go into every experiment with the assumption that it could become production ready. Even if we ultimately retire the idea, that mindset forces us to think through integration, data quality, and scalability from the start.
One thing our pilots share in common is clear intent on outcome. There’s always a North Star we’re aiming towards that will transform the business.
The key to AI success is treating experimentation as an operating model with outcome at the heart of every project. Pilots need production-level rigor; that discipline is what separates genuine learning from wasted effort.
Fail Fast, but Make Failure Inform the Next Step
When AI first emerged—especially large language models (LLMs)—the number of platforms and tools was overwhelming. We tried dozens of them. Our CEO, TVN Reddy, made it clear: If we needed to experiment, we should experiment. If something doesn’t work, we move on without dwelling on sunk costs.
That permission to fail fast was critical. But it came with an expectation: Every pilot had to start with clear intent. What problem are we solving? What does success look like? Who will use this if it works?
Without those anchors, pilots drift. Teams build interesting demonstrations that never connect to real workflows. Or they chase technology for its own sake, losing sight of the strategic purposes.
“The key to AI success is treating experimentation as an operating model with outcome at the heart of every project."
Pick Use Cases That Prove Value
One mistake we’ve seen (and occasionally made ourselves) is choosing pilots based on what’s easy to build rather than what’s valuable to prove. Those pilots might succeed technically, but they don’t build momentum. If you save one person an hour annually, no one cares. The win is too small to create belief.
At Aptean, we now prioritize pilots that affected workflows our teams used daily: support cases handled hundreds of times a day, reporting processes that consume hours each week, decision points that create bottlenecks. When a pilot succeeds in those contexts, the results make a difference. Teams start asking, “What else can we do with this?”, and that curiosity drives adoption.
Engage the Right People at the Right Time
Pilots need two kinds of involvement: the team building the solution and the people who will ultimately use it. We’ve learned to be deliberate about when we bring in each group.
Involve users too early and you risk disenchantment. But wait too long, and you miss critical feedback about how the tool fits into real workflows, as well as edge cases you haven’t considered.
The balance we’ve found is to build the first version, refine it until it’s reliable, then bring in a small group of early adopters—people who understand they’re testing something new and are willing to provide honest feedback. Their insights should shape the final iteration before broader rollout.
Key Plays:
» Start with clear intent. Every pilot should answer a specific question or solve a defined problem.
» Fail fast but learn from it. Unsuccessful pilots aren’t wasting effort; they’re data that informs the next iteration.
» Pick problems worth solving. Focus on projects that will deliver significant value to your organization, not what’s easy to build.
» Avoid premature exposure. Build reliability first, then bring in feedback to refine AI pilots before scaling.
Make Speed Your Strategy
Accept That AI Won’t Always Be Perfect
AI is evolving faster than any software category in recent memory. New models, new platforms, new capabilities, and all of it shifting constantly. If your organization waits for certainty before acting, you’ll always be reacting to change rather than leading it.
One of the most interesting shifts with AI is tolerance for imperfection. In traditional software, users expect precision. A bug is a failure. With AI, people understand that results won’t always be 100% accurate— and in many contexts, they’re fine with that. If an AI assistant produces useful insights 80% of the time, that’s still a significant improvement.
In Aptean’s early AI journey, we wanted to wait until we had all the answers; until every feature was polished and we were certain the technology would perform exactly as planned. Meanwhile, competitors were announcing capabilities that weren’t even ready yet and getting credit for being first.
Moving fast doesn’t mean shipping broken software or overpromising what you can deliver. It means being willing to release capabilities that are good enough, even if they’re not perfect. Every AI system involves trade-offs: speed versus accuracy, automation versus control, flexibility versus consistency. As a CTO, your job is to own those trade-offs.
Just remember that tolerance isn’t universal. If you’re using AI to draft an email, occasional imperfection is fine. If you’re using it to calculate financial compliance, the bar is much higher.
Remove the Distance Between Insight and Action
You can’t make your organization move faster by emphasizing urgency or setting aggressive deadlines. Speed comes from removing friction: shortening the distance between insight and action, between pilot and production, between “we should try this” and “it’s live.”
The biggest bottleneck is red tape. In most companies, moving from an idea to implementation means navigating multiple approvals, competing priorities, and unclear ownership. We’ve worked on all three areas to support rapid innovation.
One of the reasons we’re able to move quickly is that our AI infrastructure—Aptean Intelligence —is designed to work within your business context. Teams don’t need months of setup or complex integrations to start using it. They can move directly from insight to action because our AI understands their challenges and needs.
“Every AI system involves trade-offs. As a CTO, your job is to own those trade-offs.”
We’ve also focused on designing systems where iteration is the default. Improvements happen continuously, not in big-batch deployments that require months of preparation. That’s how we’ve been able to add over 1,300 new capabilities to our solutions this year alone.
Build Speed Into Your Architecture
The systems you build today shape how fast you can move tomorrow. If your architecture is monolithic, tightly coupled, or dependent on manual workarounds, adding new capabilities will always be slow. But if you design for modularity, reusability, and automation, speed compounds over time.
At Aptean, we’ve focused on creating building blocks that teams can combine and reconfigure without starting from scratch. Agents built for one use case can be adapted for another.
Intelligent systems adapt, learning from feedback, refining responses, and improving over time without constant intervention. Aptean has embedded AI agents across our solutions that interpret live data and modify their actions. When one system detects a pattern or anomaly, it can trigger adjustments in related workflows, aligning performance with changing priorities without waiting for human direction.
Modularity and adaptability will allow your speed to scale. The first AI capability you build might take months, but the 10th will take weeks. And the 100th will take days, because you’re building on proven components.
Key Plays:
» Accept imperfection where it makes sense. AI doesn’t need to be 100% accurate to deliver value, but know where the bar for precision sits in your company.
» Remove red tape between insight and action. Collapse the distance from idea to implementation by clarifying decision rights and ownership.
» Design for continuous iteration. Build systems where improvements happen incrementally, not in delayed, highrisk releases.
» Make speed architectural. Modular, reusable components and adaptive intelligence enhance your ability to deliver new AI functionality quickly.
Own the Intelligence Layer
Define the Architecture That Connects Data, Context, and Decisions
As AI becomes embedded in every system and decision, the CTO’s role changes. You’re no longer just managing infrastructure or overseeing development teams. You’re defining how intelligence flows through the organization, how data connects, and how context is preserved.
This is the intelligence layer: connective tissue linking data, workflows, business rules, and AI agents. It’s what separates organizations that get value from AI from those that end up with a collection of disconnected tools.
We learned early on that throwing everything into a repository and hoping AI would figure it out doesn’t work. When we first built our knowledge base for AI-powered support, we pulled in every document we could find. The result was convoluted, contradictory responses.
By pulling that data out, scrubbing it, and verifying the information, we ended up using fewer documents. But the results were more accurate and consistent. The lesson: More data doesn’t always mean smarter AI; quality and coherence matter more than volume.
Add Value Through Unstructured Data
For decades, enterprise software has operated on structured data. AI changes that equation. LLMs and other AI tools can make sense of unstructured data—documents, emails, chat logs, images—in ways that were impossible before.
This opens up new sources of insight, but it also creates new risks. Unstructured data is messy. It’s inconsistent. It often contains information that’s outdated, subjective, or just plain wrong.
As a CTO, you need to understand where unstructured data adds value. At Aptean, we’ve found that combining structured and unstructured data creates the most powerful intelligence. Structured data (transactional records, operational metrics, business rules) provides the foundation. Unstructured data (customer sentiment, historical decisions, domain knowledge) adds context.
“More data doesn’t always mean smarter AI; quality and coherence matter more than volume.”
Make Intelligence a Core Leadership Discipline
As AI becomes integral to decision-making, leaders need visibility into how it influences outcomes. This is a new territory for most executive teams. They’re accustomed to reviewing reports and setting priorities. They’re not accustomed to asking, “How did the AI reach that conclusion?” or “What happens if the model gets this wrong?”
One of the criticisms of AI is that it operates as a black box. Decisions happen, but no one can explain why. CTOs need to bring the intelligence layer into leadership dialogue to explain how those decisions are made, how results are reviewed, and how accountability is preserved when intelligent automation takes over.
At Aptean, we’ve developed new ways of communicating what our AI systems are doing. It’s not enough to say an agent handled 500 support cases this week. Leaders want to know: What kinds of cases? How accurate were the responses? When did the system escalate to a human, and why?
Key Plays:
» Design the connective tissue. Create an intelligence layer that links data, workflows, business rules, and AI agents into a coherent system.
» Prioritize data quality over volume. More data doesn’t mean smarter AI. Accurate, consistent information delivers better results.
» Understand the trade-offs. Unstructured data opens new sources of insight but requires curation and verification to avoid inconsistent or outdated inputs.
» Bring the intelligence layer into leadership dialogue. Explain how AI decisions are made, how results are reviewed, and how accountability is preserved.
Scale Transformation Through Small Changes
Build Momentum With Focused Wins
Grand announcements and top-down mandates don’t drive AI transformation. It builds through momentum: small teams solving defined problems, sharing proven results, and creating the case for broader adoption.
Aptean didn’t become an AI-first company overnight. We got there through hundreds of experiments, dozens of successful pilots, and the continuous effort to embed intelligence into workflows. And we learned that transformation scales faster when it starts small and focused.
Vague goals like “explore AI” or “increase efficiency” don’t create momentum. Giving your transformation direction focuses effort and makes the next opportunity easier to identify. Small changes compound into big wins.
“You don’t need everyone on board at the start. You need a small group willing to prove what’s possible.”
Rally the Herd
In the book Switch by Chip and Dan Heath—one of my go-to references for change management—there’s a concept called “rallying the herd.” People learn from what they see others doing. When early adopters demonstrate success, the folks on the fence typically follow. The laggards either come along eventually, or they don’t.
The key to rallying the herd is identifying those early adopters: people who are intellectually curious, willing to lean into uncertainty, and comfortable being the first to try something new. You don’t need everyone on board at the start. You need a small group willing to prove what’s possible.
At Aptean, we started AI innovation with teams that had the appetite for experimentation. We gave them the tools, autonomy, and support to build AI agents for their own use cases. When they succeeded, we showcased those wins. And the folks on the fence—those who needed to see something work before they’d try it themselves—started asking questions. “How did you do that? Could we apply this to our process?”
Some people will resist change, cling to the old ways of doing things, or dismiss AI as hype. That’s fine. Your job as a CTO isn’t to convert everyone. It’s to create the conditions where those who are ready can move forward, and where success is compelling enough to bring along those who want to see results before they dive in.
Co-Create With Colleagues and Customers
One of the advantages of starting small is that successful pilots can be adapted and reused. Agents we built for one team’s workflow have become templates for other use cases. We’ve now extended this principle beyond internal teams to include customers.
Early-adopter programs, advisory boards, and industry workshops give customers a voice in shaping how AI is applied and refined. These collaborations ensure that every new capability is tested against real operational challenges.
Co-creation also builds stronger partnerships. Colleagues and customers don’t feel like they’re being sold to; they feel like they’re contributing to something meaningful. And the feedback we get from those engagements makes our solutions better for everyone.
Key Plays:
» Scale through small, focused wins. Transformation builds faster through dozens of successful pilots than through organization-wide mandates. Start with a specific problem.
» Focus on the willing. Don’t waste energy trying to convert AI resisters. Find people who are intellectually curious and willing to experiment, then give them the tools and autonomy to prove what’s possible.
» Co-create. Involve colleagues and customers in shaping how AI is applied to ensure it solves real operational challenges and nurtures deeper relationships.
Where We Go from Here
I started this playbook by admitting I don’t know what the development landscape will look like in six months. That’s still true. And it will probably be true six months from now as well.
But I’ve learned that not knowing doesn’t mean not moving.
At Aptean, we’ve made mistakes. We’ve invested time in platforms that didn’t work out. We’ve had leaders resist change and teams struggle to adapt. None of that stopped us from becoming an AI-first company.
When we launched AppCentral, we didn’t have all the answers. We deployed it internally first, learned from what broke, and built confidence through real experience. That same approach—test rigorously, learn fast, scale deliberately—has guided every AI initiative since.
Your organization will face different challenges, different constraints, and different opportunities. But the core truth remains the same for all of us: AI adoption requires courage, discipline, and the willingness to disrupt.
If you’re looking for a risk-free path forward, you won’t find one.
If you’re waiting for the perfect moment, it doesn’t exist.
But if you’re willing to become Customer Zero, to pilot relentlessly, to make speed your strategy, to own the intelligence layer, and to scale through small wins, your teams will be refining their third generation of AI agents while others are still writing their first pilot brief.
I hope this playbook gives you the confidence to begin, or the validation to keep going.
Jenny
What can I help you with?
