9 product leaders share their tips for introducing AI to your product

As AI rapidly evolves, the pressure is on for B2B SaaS companies to strategically integrate this powerful technology into their offerings. We spoke with top product leaders to get their insights on introducing AI functionality that delivers real value, avoids the hype, and keeps users engaged. Here’s what they had to say:

Erik Almenberg, CPO at Kognity

1. Value over hype: We’re past the phase where AI is new and shiny. Focus on use cases that create real value for customers, rather than just on the technology itself.
2. Customer-level opt-out: Risks are at the company level, not just at the user level. By giving customers control, you increase appeal to cautious prospects and reduce future churn from those who may become wary of AI.
3. Align revenue model with AI expenses: The typical SaaS model of tiers and users doesn’t align well with the usage costs of high-powered LLMs. The use case and type of LLM will affect costs, potentially requiring changes to pricing and packaging.

Jakob Nettelbladt, CPO at Mentimeter

1. Users aren’t looking for AI features; they’re looking for ways to solve their problems more efficiently. Offer them solutions, not just new tech.
2. With larger customers, getting them to opt-in to AI features will be your biggest barrier. Don’t underestimate the design challenge this presents.
3. AI that simply saves time will become a commodity. To have a real impact, your AI needs to not only “do the thing” for users but do it better than they could themselves.

Maja Lindström, CPO at Talentech AB

1. Security and compliance: AI brings great opportunities but also risks related to privacy and data usage. Be proactive about privacy regulations, including the upcoming AI Act. Implement robust security measures and be transparent with users about data usage to build trust. Compliance can be a differentiator when everyone has access to the same tools.
2. User-centric development: To truly understand how AI will impact user experience and create value, conduct thorough user research before, during, and after AI implementation. AI functionality should solve real needs, not just be a buzzword. Address user fears and hesitations to ensure adoption.
3. Bias mitigation and transparency: AI can reproduce bias, so it’s crucial to train models on diverse data and continuously test outcomes. Be transparent with users about the possibility of mistakes and manage expectations accordingly.

Vincent Jong, VP product at Dealfront

1. Leverage unique data: Everyone has access to the same APIs. To build something truly differentiating, use unique data. Simple use cases might seem cool at first, but they’ll quickly become commoditized if they aren’t already.
2. Focus on adoption: Usage-based pricing models with free tiers and additional AI credits are gaining traction. Therefore, ease of use and adoption are key to ensuring that users experience value and want more.
3. Privacy considerations: Most AI features rely on external services like ChatGPT. Avoid sending identifiable or sensitive data to these services, as it could become a hurdle during sales negotiations, particularly with GDPR compliance.

Dimitra Retsina, Strategic Product & Technology Advisor at Kelome

1.⁠ ⁠Development effort requires a different approach: Developing AI functionality is inherently non-deterministic, which requires a focus on quality assessment and continuous improvement.

2.⁠ ⁠Human factor for quality control: Significant resources are needed to assess and safeguard the quality of AI outputs.

3.⁠ ⁠Focus on what instead of how in solving a user problem: Focus on solving the problem your product addresses rather than obsess with the type of solution and keep ethics, privacy, and compliance considerations in mind

Peter Sunna, CPO at Kognic

1. Solve real problems: Help users in a way that truly improves their experience. If you’re branding it as “AI something,” you’re likely chasing a trend instead of solving a genuine problem in a new way.
2. Be honest: Users need to understand how AI works and how it impacts their actions. If they don’t trust it, they won’t use it.
3. Design for scale: Ensure your AI feature works as well, if not better, as more people use it and as more data is added.

Sigrún Gunnhildardóttir, CPO at AGR

1. Set realistic expectations and be transparent: AI can significantly enhance products, but don’t overpromise or rely on AI to solve every problem. Transparency is key—if AI isn’t the right tool for a task, make sure users know that and feel in control. Overhyping AI can lead to frustration and loss of trust.
2. Leverage AI strategically: Use AI where it excels, such as in pattern recognition, automation, or predictions. Complement it with traditional methods for tasks where AI might not be the best fit. A balanced approach ensures your product is both effective and reliable.
3. Prioritize user experience with AI: Use AI to enhance user experience, especially through generative AI. Users expect intuitive interfaces where they can “talk to” software rather than navigate complex systems. This can be a quick win when implemented correctly.

Davina Erasmus, VP Product at Crunchr

1. Control: Ensure that what you allow others to use is within your scope, keeping the focus on the outcomes you’re targeting.
2. Context: Avoid black-box approaches by providing touchpoints that let users engage with the process and take ownership of the outcomes.
3. Capability building: Provide users with outcomes that guide them on how to take the next step in the conversation.

Heta Pirttijärvi, VP Product at Sievo

1. Think about what you develop yourself vs. what you buy or connect with, depending on what your core business is (e.g., why build a chatbot from scratch when you can utilize OpenAI developments, etc.).
2. Think about the communication clearly. AI is still seen in different ways, with some being allergic to AI without human supervision and some to the ‘powered by AI’ fluff. What is your audience, and how do they think of the topic?
3. Think about the real use case and customer value. There’s a lot of problems that don’t need AI and that are better resolved with another solution, too.
4. Bonus: Ensure you have a good team that keeps up with the new tech advancements and developments. This field is developing fast, so staying up to date is crucial!