Key Takeaways:
|
In the latest episode of the Spitfire podcast, Kgomotso Dibetso speaks to Brandon Holz, Spitfire Inbound’s Head of Implementation and AI Development, about what businesses need before AI becomes useful in revenue operations.
Brandon gives us great starting points and shares his experience with implementing AI. His main point is that AI needs the right conditions to work properly: a clear problem, reliable data, people who know how to use the tools and enough governance to keep experimentation helpful and secure.
Q: Why is it important to start with a specific business problem rather than just rushing to adopt AI tools?
A: Many organisations make the mistake of adopting AI platforms simply because they are trendy, leading to a flurry of activity that doesn't actually translate into meaningful business progress. To avoid this, businesses must treat AI like any other specialized tool, using it only when it’s the right fit for the task. As Brandon Holz explains, "AI is a tool like any other tool." Just as a handyman doesn't use the same tool for every job, RevOps teams need to focus on context and specific use cases. By prioritising practical business problems over the pressure to "use AI," teams can ensure they are deploying technology that actually solves a challenge rather than just adding complexity.
Q: How does data quality directly impact the effectiveness of AI in a RevOps strategy?
A: The reliability of any AI output is strictly bound by the quality of the information it is fed. If a CRM is cluttered with "dirty data", such as duplicate companies, vague deal stages, missing contact roles, or inconsistently applied lifecycle stages. The AI’s output will be similarly compromised. Brandon emphasises that the results are only as good as the input: "It’s only going to be as good as the inputs that you give it." Consequently, before integrating AI into revenue processes, a CRM data strategy is non-negotiable. Organisations must identify which fields are reliable, clarify data ownership, and define protocols for correcting data errors to ensure the AI has a foundation it can actually trust.
Q: Why is the "human factor" the most critical element of AI readiness?
A: While AI discussions often veer into technical details like models and prompts, the ultimate success of these tools depends on the people using them. Brandon argues that technology alone isn't the solution; the impact of that technology is determined by how well the human team engages with it. He notes: "The more impact you want your AI and your technology in your business to have, the more you need to actually focus on the human beings." Successful AI enablement requires problem-solvers who can communicate effectively with the technology; explaining the problem clearly, providing context, and applying human judgment to the outputs before they are finalised. When AI tasks are aligned with an employee's daily responsibilities, like summarising tickets for service managers or drafting nurture variations for marketing, the value becomes obvious, making adoption much more natural.
Q: How can businesses effectively encourage widespread AI adoption across different roles?
A: To foster a hunger for AI, businesses should focus on demonstrating how the technology creates convenience and eliminates repetitive administrative work. Rather than enforcing adoption, companies can create a culture of experimentation through initiatives like "AI Wednesday" sessions, which provide a safe space to share real-world use cases, review what failed, and brainstorm new ideas. Furthermore, identifying "AI champions", employees who are already testing tools and asking the right questions, is vital. However, because AI impacts so many workflows across an organisation, it is a team effort rather than a one-person job. The key is making the benefit obvious in people’s day-to-day work.
Q: What is the most effective approach to AI governance in a business?
A: Governance is a critical business necessity, especially since many employees are already experimenting with AI tools independently. To regain visibility and manage risk, Brandon suggests forming an AI governance council composed of key stakeholders from across the business. This group should be responsible for reviewing proposed use cases and determining the appropriate levels of risk versus reward. A crucial first step in this process is establishing a mechanism for transparency: "Make sure that you’ve got some way of tracking what AI is being used in your business and how it’s being used." By allowing employees to submit tools for review, the business can classify risks accordingly; treating high-stakes, customer-facing AI applications with more rigor than low-risk, internal brainstorming tools, while leveraging existing security and data protection frameworks like GDPR and POPIA.
Q: What is the ultimate takeaway for RevOps teams regarding AI readiness?
A: For RevOps, AI is not a magic fix for broken operations; it acts as a force multiplier for a system that is already organised. Before implementing any AI tool, RevOps teams should conduct a health check of their existing infrastructure; including lead movement, data capture, reporting structures, and team trust in the system. AI should only be introduced when there is a clear problem to solve, sufficient data to support the process, and a team trained in responsible use. Ultimately, a successful RevOps AI strategy rests on four pillars: a defined problem, clean data, empowered people, and established rules of governance.
Many AI conversations start in the wrong place. New tools or platforms are created rapidly, and teams might start using them before considering what they can actually change. The result is activity, not progress.
Brandon puts it simply:
“AI is a tool like any other tool.”
That framing centres around use cases and context. AI may be powerful, but it still needs to be used in the right scenarios. Brandon compares it to a toolbox. If you are a handyman, you don’t use every tool for every job. You choose the one that fits the task.
For a RevOps team, AI conversations should start with practical questions:
Brandon kept coming back to data quality.
He explains:
“It’s only going to be as good as the inputs that you give it.”
If your CRM has duplicate companies, vague deal stages, missing contact roles or lifecycle stages nobody follows consistently, AI has to work with that version of data; dirty data.
CRM data strategy needs to be part of the AI conversation from the beginning. Before AI is added to a revenue process, you need to know which records matter, which fields are reliable, who owns the data, and what happens when the data is wrong.
AI discussions run the risk of becoming very technical very quickly. We start talking about models, prompts, tools, and outputs.
Brandon’s view is that the biggest success factor is the humans using the tools, not the tools themselves.
As he says:
“The more impact you want your AI and your technology in your business to have, the more you need to actually focus on the human beings.”
Brandon talks about the value of people who think like problem solvers. He also points to communication as a practical AI skill. With large language models, poor instructions usually lead to poor outputs. If a user can’t explain the problem, provide useful context, or check whether the output makes sense, the tool will only get them so far.
That makes AI enablement a people matter as much as a systems matter.
A salesperson might want help preparing for a meeting using CRM notes and company context. A service manager might want to summarise ticket themes. A marketing team might use AI to analyse campaign performance or draft variations for a nurture sequence, then apply human judgement before anything goes live.
The closer the use case is to someone’s daily work, the easier it becomes for them to understand why AI matters and the value it can provide.
Q: How can businesses create hunger for AI adoption?
A: Show people the convenience. People adopt tools when the benefit is obvious in their day. If AI saves them from doing admin or redoing work multiple times, the value is easier to communicate.
Brandon mentions our AI Wednesday sessions as one way the team keeps the conversation going. It’s a platform to share real use cases, show what worked, explain what did not, and give people ideas they can actually try.
Brandon also recommends identifying AI champions across the business. These are the people already testing tools, asking better questions, and finding useful applications inside their teams.
But he’s clear that one person can’t carry adoption for the whole organisation. AI touches too many roles and too many workflows.
When Brandon touches on governance, he highlights that in many businesses, AI use is happening before a formal process exists.
Employees may already be testing AI tools for research, writing, reporting, admin or customer-facing work. Not every experimental use case carries the same level of risk, but the business still needs visibility.
Brandon recommends forming an AI governance council made up of stakeholders from various key functions in the business. This group can review proposed use cases, assess the tools being used, and determine the level of risk (and reward) involved.
One of the first steps, he explains, is to:
“Make sure that you’ve got some way of tracking what AI is being used in your business and how it’s being used.”
A practical starting point is to create a way for employees to submit AI use cases and tools for review. From there, the business can classify risk and route higher-risk use cases to the right people.
A low-risk internal brainstorming use case will not need the same level of scrutiny as a tool that processes customer data or affects customer-facing communication.
Brandon also points to existing data protection and security frameworks such as ISO, the EU AI Act, GDPR, and POPIA. The business doesn’t need to invent every principle from scratch. Much of the guidance around risk, privacy, and responsible use already exists. It’s just about applying it to daily operations.
The aim is to keep useful experimentation moving while making sure sensitive use cases are reviewed properly.
For RevOps teams, AI needs the business to be organised enough for the output to mean something. Before asking which AI tool to use, businesses should look at how leads move through the CRM, how data is captured, how handovers happen, how reporting is built, and whether teams trust the system enough to use it to its full potential.
Brandon is clear that AI should be used only when it solves a real problem, when the data is sufficient to support it, and when people understand how to use it responsibly.
AI can support RevOps, but it needs a solid foundation. That starts with the problem, the data, the people, and the rules of use.