RAF

Data, Risk & Growth in Practice

Artificial intelligence (AI) is no longer a future-state initiative. For mid-market companies, it has quickly become a lever to drive operational efficiencies.

However, scaling AI without structure introduces complexity and risk. The challenge is not whether to adopt it, but how to do so responsibly, efficiently, and with measurable impact. That starts with how applications like Claude or ChatGPT are introduced into workflows.

Earlier this year, we were joined by digital marketing experts from Seer Interactive. Wil Reynolds, CEO & Vice President, and Alisa Scharf, VP of AI & Innovation, shared insights during our January AI Summit that highlight how this is playing out in practice.

The Hidden Risks of Deploying Artificial Intelligence

Before deploying AI, a few key considerations should be addressed. Three risks tend to surface early, often before teams have clearly defined how it should be used.

• First, the longer you take to deploy, the greater the risk of Shadow AI activity. Essentially, AI-forward team members will use this technology whether it’s sanctioned or not. And if they’re using the free tools, you’re at risk of data leakage into training models.
• Second, some of the most compelling use cases of AI involve the ability to both read from and write to other tools and platforms. This is where the magic happens, but also where risk can build up if you aren’t testing and selecting the right tools and frameworks.
• Third, you need to have enablement figured out before you hit go. Assess what training you’ll provide (AI readiness), the use cases you’ll support, and how well the technology integrates with your existing stack. For example, you can’t easily get your team on Claude Cowork without a little IT admin support. And perhaps most importantly, do you have a clear policy for what to do and what not to do? All of these things must be thought through before you fully deploy.

Unsanctioned AI usage can quietly introduce data leakage, compliance exposure, and fragmented workflows. At scale, these risks can compound across companies if structure isn’t in place.

Safeguards Founders Should Consider

Successful AI adoption is not about broad rollout; it’s about controlled implementation. Having a dedicated enablement team is critical here. Roll out in batches or tiers to ensure the tooling works as expected and the use cases are clearly defined and operational.

Equally important is aligning AI with how work actually gets done. Key questions to ask yourself:

• How is work done today, and how do new AI tools impact those workflows?
• Is it clear what individuals must start doing?
• How will you measure ROI? It’s ideal to have a measurement strategy. Keep it simple and directly attributable. Establish a clear benchmark of current-state performance to accurately measure efficiency over time.

Lastly, you need executive modeling to be successful. Change is hard, and it’s especially hard when you’re an IC who doesn’t have visibility into all decisions made about what, why, and how. Show your team you’re embracing the chaos of change with them, transparently. This is an exciting venture, but it’s also overwhelming. Show them you’re in it together.

How Data Maturity Influences AI Integration

AI performance is directly tied to the quality and accessibility of data, but data maturity goes beyond structured databases.

We’re used to thinking about data as numbers in a database, but for an AI-forward organization, data is everything. This includes everything from workflow documentation to call transcripts with clients.

Starting with an audit is always a good step. Assess all the data that goes into the work being produced, as well as its current state. That includes how well it’s structured, whether pipelines are established, or if manual action is required to transform and relocate data.

By concurrently performing a Disruption Analysis, you can begin to identify your highest-priority opportunities and use cases, and use that direction to isolate the data governance steps that need to happen. Don’t try to assemble your 12-month roadmap as part of this exercise.

The technology is changing too rapidly. Focus on where you can make a meaningful impact, whether that’s in workflow deployment or foundational data governance.

Three Practical Tips For Leveraging AI Today

For organizations looking to move from strategy to execution, progress often starts small. Wil and Alisa share practical ways to get started:

• Pick a workflow you personally do every week and work with a state-of-the-art model like Claude Opus or ChatGPT 5.4 as a thought partner to rebuild that process end-to-end. You may not succeed, but you’ll learn invaluable insights that will guide the next build.
• Share your wins and losses with your team. The more we share, the more we learn. Make it clear that it’s okay to fail and get it wrong, and ensure those painful lessons are learned by everyone.
• Don’t roll out new workflows or tools without a clear directive on what the team must start doing and what it must stop doing. Change management has long been underappreciated in knowledge work, but teams with an established process will be the ones to get ahead.

AI as a Driver of Portfolio Growth

AI is no longer experimental–it’s foundational. The companies that win will not be those that adopt it fastest, but those that adopt it most deliberately.

Richard Deede, Director of Digital Marketing, shares how AI has become an extension of their team, and is being implemented across the portfolio.

“As a PE firm, we utilize AI for data analysis at scale to gather insights and establish an alerting system that highlights landscape changes, shifts in traffic and buying behaviors, and other red flags for further investigation. It’s trained to understand which KPIs are critical for each individual portfolio company.”

Understanding how AI operates within the conditions and instructions you provide helps determine what works and what doesn’t. From there, iteration allows for incremental improvements that can be tested, refined, or rolled back as needed. He adds:

“The safeguards we prioritize are centered on data privacy and ensuring we don’t expose sensitive financial or customer information. Our rule of thumb is that if the info is already publicly available, it’s okay to share it with our AI agents. If there is a grey area, our recommendation is to anonymize the data before granting AI access.”

One of AI’s strengths is its ability to harmonize data across platforms, even when sources differ significantly (ie, session data from GA4 vs. Shopify). “The more data we can feed in, the stronger the confidence we have in the insights coming from the output.”

For firms operating across multiple companies, this balance isn’t optional. It’s what determines whether AI creates value or introduces risk at scale.

Learn More About RAF

Private Equity Firm

Founded over 40 years ago, RAF acquires positions in middle-market companies across a diverse set of industries. The firm maintains a long-term strategy focused on strong management teams and a potential for sustained, accelerated growth. Contact us to reach a member of our acquisition team.