Dialpad’s Top Learnings Building Its Own AI Stack
When it comes to AI implementation in SaaS, most companies are still figuring out whether to build or buy. But Dialpad has been quietly building their AI advantage for over 7 years, processing more than 8 billion minutes of conversations and hitting $300M ARR along the way.
At SaaStr AI Day 2025, Jim Palmer, Dialpad’s Chief AI Officer (and co-founder of TalkIQ, which Dialpad acquired), shared the tactical playbook they used to build and scale their AI capabilities. Here are the key learnings every SaaS company should know.
1. The Build vs. Buy Decision Isn’t Binary
A common mistake companies make is thinking they need to either build everything from scratch or rely entirely on third-party APIs. The reality is more nuanced:
- Start with clear use cases: Don’t implement AI just because it’s trendy. Dialpad began with specific problems they wanted to solve around conversation intelligence.
- Understand the trade-offs: Real-time processing has different cost, throughput, and risk profiles compared to batch processing. Make this decision early.
- Consider a hybrid approach: You can start with third-party APIs while building your data foundation, then gradually move to custom solutions as you scale.
2. Domain Data Is Your Secret Weapon
Dialpad’s massive advantage comes from their 8 billion minutes of processed conversations. Here’s why domain data matters:
- Tighter accuracy gains: Domain-specific data allows you to optimize for specific use cases while maintaining general capabilities
- Vertical specialization: Different industries have different needs – Dialpad can tune their models for sales teams vs. support teams
- Competitive moat: While anyone can access public AI models, your domain data is unique to your business
3. Data Governance Needs to Start on Day One
Palmer emphasized that data governance isn’t just about compliance – it’s about building a sustainable AI advantage:
- Own your data story: Even if you’re using third-party APIs, start collecting and organizing your data immediately
- Build responsible AI practices: Implement clear policies about data usage, privacy, and model training
- Plan for different localities: Design your data architecture to handle different regulatory requirements (EU vs. US)
4. The ROI of AI Needs to Be Measurable
Dialpad’s success comes from relentlessly focusing on customer value:
- Customer success metrics: One customer achieved 20% efficiency gains in workflow improvement
- Specific use cases: AI Recaps feature distills key information from calls in real-time
- Continuous measurement: Use telemetry and observability to track accuracy and impact
5. AI Should Augment, Not Replace
A key insight from Dialpad’s journey is that AI works best when it enhances human capabilities:
- Focus on assistance: Their AI coaching features help sales and support teams improve, rather than replace them
- Handle simple tasks: AI can take over routine queries, freeing humans for complex problems
- Measure human impact: Track how AI tools improve human agent performance and satisfaction
What’s Next?
Looking ahead, Palmer shared some interesting predictions:
- Complex voice communication in contact centers won’t be fully automated in the next 24 months
- The focus will be on building better tools to assist human agents
- The combination of real and synthetic data will be crucial for future AI development
The Bottom Line: It’s a Marathon, Not a Sprint
Dialpad’s journey to $300M ARR through AI wasn’t overnight. They:
- Started with clear use cases
- Built a strong data foundation
- Focused on measurable customer value
- Invested consistently over time
- Maintained a human-centric approach
For SaaS companies looking to leverage AI, the key is to start building your data advantage now, even if you’re using third-party tools. Your unique data will be your competitive advantage in the AI-first future.
