Executive Summary
GenAI’s potential is massive – but it’s definitely not “plug and play”, and beneath the hype sits real complexity… Your data must be “GenAI Ready”, your people will need a wide range of new skills, and your dreams of Agentic AI-enabled Intelligent Automation need to be approached with care. In addition, if you can’t explain, with metrics, where GenAI is helping and where it isn’t, your insights aren’t ready either.
While adopting and scaling GenAI isn’t simple, the benefits are clear and compelling. You’ll need to take a structured, systematic approach, though, to “Win with GenAI”. Start by understanding what’s required to adopt and scale GenAI, then treat it as a “strategic imperative”, as it’s how the next decade of growth will be decided…
“Who Knew AI is Still Hard?”
Somewhere along the GenAI Hype cycle, many folks have lost sight of the “AI” part of Generative AI… Slogans like “If you can think it, AI can Build It!” are everywhere, and many of our non-technical brethren are nursing dreams of building full-stack, high-performing apps over lunch…
The truth is GenAI’s potential is enormous, but its current state capabilities are still maturing. Not a surprise given where we are in GenAI’s innovation and adoption cycle, but it’s important to understand and solve for the underlying complexity as companies work to embed GenAI into their core ways of working. This is particularly true of the organizational capabilities required to successfully adopt GenAI @ scale. Here are some targeted “lessons from the trenches” …
Your Data Isn’t Ready
This isn’t conjecture, it’s a fact… GenAI significantly raises the bar in terms of Data Maturity expectations. In traditional application development, we define what functionality will be in-scope, what data will be needed, and structured ways to clean and prepare our data to meet those defined needs. For my technical friends, think of the range of ETL jobs that are required to feed your applications.
GenAI users, on the other hand, are given the ability to ask a nearly infinite range of unstructured, natural language questions that your data needs to be able to answer, in real-time. That means your data needs to be trustworthy, discoverable, understandable, connectable, and highly performant. Few organizations can claim with a straight face that their data meets those criteria today – but that will be the “price of admission” for leveraging GenAI moving forward.
Your Talent Isn’t Ready
Few organizations have an army of AI Engineers to drive their GenAI Vision. Unless you’re Google or Meta, you’re likely going to significantly increase your AI Engineering talent pool over the next 2 – 3 years. Most companies turn to their favorite HR partner and ask them to lead a recruitment effort, only to find that every other company is searching for the same talent. In parallel, organizations are working to “upskill” their existing team and accelerate their GenAI technical development. While this is the right approach, the journey from traditional application development to non-deterministic AI Engineering is a complex one, and purchasing a license to LinkedIn Learning or Coursera isn’t going to drive the outcomes you’re looking for @ scale.
This isn’t just an Engineering / App Dev issue either. Product Managers, BA’s, Marketers, Operations, Compliance & Legal, and HR team members will all have their own journey to work up the GenAI learning curve. The bottom line – an integrated approach to GenAI Talent Development will be needed for organizations to “Win” with GenAI.
Your Agents are Definitely Not Ready
Agentic AI will transform many aspects of how organizations deliver value and operate their business. That being said, there is significant work to be done to transition that Vision into reality. The data challenges noted above must be solved, “tribal knowledge” needs to be codified, overly complex business processes need to be optimized, and rigorous guardrails need to be in place.
Indy Sawhney, a GenAI thought leader at AWS, captured this reality really well in his “Weekly Dose of GenAI” newsletter where he advocates starting small, and iteratively building towards autonomous Agentic capabilities. I particularly like the focus on “SUV’s” – not the large 4-wheel drive version, but an approach that focuses on introducing the “Smallest Unit of Value” with agents, and then building a clear path to higher impact. Check out his blog here – it’s worth a read.
Your Diagnostics (probably) aren’t Ready
“Friends don’t let friends build ‘black box’ GenAI solutions”… Most of us had a similar experience when we first began working with LLM’s and building GenAI solutions – progress was largely based on “trial and error” and “tinkering”. For many of us, that was part of the fun… When we were asked to provide production-quality solutions that consistently delivered predictable, high-quality outputs, it was a different story…
Few organizations today have adopted Evaluation Driven Development (EDD) practices across their GenAI development efforts, but they need to. This isn’t a new concept, as Data Scientists have been defining clear Experiment & Model Success Specifications, such as F1 scores, Predictability, and Drift measures, for years. Product Managers and developers need to take a page from the Data Scientist playbook and incorporate EDD-practices into their core Software Development Lifecycle (SDLC) processes to ensure proper transparency, accelerate velocity, and improve solution quality.
Your Insights aren’t Ready
If you ran into someone from your Board in the elevator, and they asked you where your GenAI efforts were having an impact, where they weren’t, and why, would you be able to answer them in a data-driven way? If not, your GenAI Insights aren’t “ready for primetime”.
If you’re serious about scaling GenAI, you need to prioritize adopting clear measures of success across your in-scope efforts. Areas of focus should include:
- Your Core Business Measures (KPIs)
- Customer & Products Metrics
- OKRs & KRIs to measure your Strategy Execution and Risk Management
- Data & Talent Readiness measures
- Detailed Diagnostics and Maturity measures
Once you’ve defined your integrated measures of success, you’ll need to turn a mountain of data into clear, actionable insights. While this has often been treated as discretionary or a “nice to have”, once you adopt semi or fully autonomous GenAI functionality across your business, integrated insights become non-negotiable.
5 Actionable Steps to Accelerate Your GenAI Readiness
1. Prioritize a Systematic Approach to Adopting & Scaling GenAI
This is the single most important recommendation I can make for leaders working to adopt and scale GenAI… If you solve “one problem at a time”, you’re looking at an extremely long lead time to value with GenAI. This is complex stuff, with a broader set of interdependencies than anything we’ve seen before. Make sure you’re taking a holistic approach to solving your GenAI needs, or you’ll be surrounded by “PoC Zombies” and playing a never-ending game of “Whack-a-Mole” across your GenAI solutions.
2. Leverage Targeted Assessments to Understand Your Baseline Readiness
Speed and focus will matter when adopting GenAI. Start by understanding what good looks like, where you’re ready, and where you’re not. Structured GenAI assessments are critical for developing a clear understanding of your current state readiness and developing targeted plans to accelerate your gap closure efforts. Reach out if you’d like to explore Accelerated Innovation’s Keys to Winning GenAI Diagnostic Framework, as we’ve spent 1,000’s of hours building an integrated, progressive assessment framework to help you compress your time to GenAI value.
3. Don’t Reinvent the Wheel, but Customize Where You Need To
Early adopters were forced to take an organic approach to experiment with and adopting GenAI – you don’t need to… Leverage a proven GenAI adoption and scaling approach that addresses the various capabilities you’ll need, in the order you’ll need to solve them. There’s no “one size fits all” approach to Winning with GenAI, though, so you’ll also need to customize your plan to meet your specific business needs. The best GenAI adoption and scaling approach is based on clear lessons learned elsewhere, with the flexibility to meet your specific use cases.
4. Prioritize Quick Wins While Your Close Capability Gaps
While you’re closing foundational capability gaps, it can be easy to underinvest in targeted “Quick Wins”. This is a mistake that several folks I’ve worked with have made, and have come to regret… Prioritize delivering clear “reasons to believe” at least monthly (and preferably more frequently) to build and sustain momentum for what will be a sustained effort. Reach out if you’d like to explore our “Quick Wins Catalog”, as we’ve compiled > 100 targeted scope solutions that can be quickly implemented as clear proof points along your GenAI journey.
5. Approach Adopting GenAI as a “Make or Break Moment”
This last point may be more controversial, but you need to approach adopting GenAI as a strategic imperative. When Web Commerce and internet-enabled capabilities hit in the late 90’s, early adopters recognized their implications and worked through the early “messiness”. Organizations that successfully incorporated those capabilities realized significant benefits, while late adopters struggled – or worse (think Blockbuster…).
GenAI should be viewed in a similar light as a disruptive technology that will have significant implications for the majority of industries and functions. Those that “lean in” and make it their own will reap significant benefits, but organizations that take a “wait and see” approach are likely to struggle to catch up if they fall significantly behind.
