Part 2 of the Systemic Shift to Scaling AI from Pilot to Scale series, examining how organizations can transform AI initiatives into enterprise-wide value
The Recipe for Real Scale In our last post, we talked about the mental shifts you need to move your organization out of the AI pilot pit and start driving real value. We talked about moving from “tools” to “systems” and why speed matters.
Now, how do we get practical? First of all we need to discuss Phase 1 which we call, Setting the Foundations.
Before you turn on out of the box features, like Agentforce in Sales Cloud, or Now Assist in ServiceNow, you need to make sure your organization can actually handle them. Think of this not as a technical checklist, but as a recipe to make sure you don’t spend millions on your new Ferrari only to be stuck driving it on a dirt road.
Here are the five steps to get your enterprise ready.
Step 1: Be Honest About Where You Are
The Reality: Most companies think they are either “behind” or “ahead.” The truth is usually messier. Your readiness is likely fragmented and uneven.
The Fix: Your Cloud Infrastructure team might be ready to ship AI tomorrow, but your Sales Ops team might still be struggling with basic data entry and salesforce data quality issues. That is normal.
As leaders, you can’t use a “one-size-fits-all” strategy. You need to identify which teams are ready to run and which ones need help walking. For example, you can use Salesforce Data 360 (formerly Data Cloud) to bridge these gaps, helping the slower teams catch up by giving them access to the same clean data the fast teams are using. It might not give you the real time insights you were promised out of the box, but it will make the user a 100x better experience if agentforce is grounded on a clean and trusted data source from day one.
Questions to ask yourself:
- Which of your departments are already using AI on a daily basis (Shadow AI)?
- Is your data team speaking a different language than your Salesforce or ServiceNow team?
- Is revenue the same thing as total sales ? If your people are confused by these types of questions, your AI solutions will also be confused.
- Do you know which teams are actually ready, and which ones are just pretending?
Step 2: Executives Need to Listen, Not Just Talk
The Reality: Executives often think their job is just to sign the check and mandate the tool. That usually backfires.
The Fix: We often see executives get super excited about AI and accidentally burn out their teams by throwing too many new toys at them at once.
To really make a tool like Agentforce or Now Assist work, you need everyone on board. Sure, leadership needs to set the big picture, but you absolutely have to listen to the people on the front lines, too. They’re the ones who know exactly where the data is messy or where the processes are broken. If you ignore how things actually work day-to-day, that “strategic vision” is going to crash and burn when you try to roll it out. The biggest thing to remember? AI won’t magically fix bad processes and garbage data, it’ll just make the problems worse.
Questions to ask yourself:
- Is your leadership actually using these AI tools yourself, or just reading reports about them?
- Do your employees feel safe telling you when an AI idea is bad?
- Do you know the difference between a cool demo and a working product?
Step 3: Fix Your Data (Seriously)
The Reality: At its core AI is just a prediction engine. If you feed it bad data, it will give you bad predictions, confidently.
The Fix: This is usually the hardest part, but you can’t skip it. Tools like Agentforce are only as smart as the data it can see. If your customer data is trapped in five different silos, your AI agents are going to look awfully in front of your customers.
In the Salesforce ecosystem, we use tools like Data 360 (formerly Data Cloud) to act as a “translator,” connecting the messy data in your backend systems (like AWS or Google Cloud) with your CRM. It doesn’t have to be perfect overnight, but it has to be connected. Treat your data foundation like a living thing, you have to keep feeding it and cleaning it.
Questions to ask yourself:
- Is your data strategy just fixing typos, or is it designing for AI?
- Have you actually connected your different data systems, or are they still islands?
- Can your AI read the “unstructured” stuff (PDFs, emails) that actually explains why a deal closed?
Step 4: Governance Lets You Go Fast
The Reality: In our experience, people get a bit spooked by the word “governance.” They think it means “red tape” and “slow down.”
The Fix: Think about a race car. You don’t put giant brakes on a Formula 1 car so it can drive slowly; you put them on so the driver feels safe driving fast.
The same applies here. We need “Governance for Motion.” If your team knows exactly what data is safe to use and that they can trust the answer they are getting from AI, they will adopt it faster. If they are scared they might leak data or respond to a customer with bad information, they will freeze and revert back to the old process. Good governance speeds things up because it removes the fear.
Questions to ask yourself:
- Does your approval process take weeks (too slow) or days (just right)?
- Do your employees trust that the AI is secure and accessing clean trustworthy data?
- Is your rulebook simple enough for a normal person to understand? If it’s not, AI is not going to magically understand it for them.
Step 5: Pay People for the Right Outcomes
The Reality: If you measure people on “hours worked,” they will hate AI because it reduces their hours.
The Fix: You have to change the goalposts. Align your incentives with Compounding ROI.
Stop measuring vanity metrics like “how many people logged in.” Start measuring business results: Did we close deals faster? Did customer satisfaction go up? Did we free up staff to do higher-value work? If you want your team to use AI, incentivize them to be efficient, not just busy.
Questions to ask yourself:
- Do your compensation plans reward efficiency, or do they accidentally punish it?
- Does everyone agree on what “Success” looks like? Does everyone know and align on how core KPIs are calculated?
- Are you measuring results, or just the effort?
Next Steps
If you get these five things right, being honest about maturity, listening to your team, fixing the data, setting safety rails, and aligning incentives, you are ready to scale. I know this sounds like a lot and that it touches so many areas, but AI at enterprise is just that, a transformational technology for your entire business. Without these core foundational pieces set, AI can quickly become a hindrance and performance bottleneck instead of the promised efficiency booster it should be.
In our next post, we’ll talk about Phase 2: Creating AI Fluency, and how to teach your team to actually drive this car you’re building.
If you’re not sure if your data is ready, connect with us. We can look under the hood and help you get moving in the right direction.
