How can founders use AI to increase profitability?

Google for Startups: What’s the most tedious task that AI has automated for you? 

Mirzaian: Our AI-powered Life Insurance Assistant significantly reduces the friction of filling out forms and answering user questions. The AI can fill out our life insurance form (via our API) with a single sentence and customers can also ask questions about coverage and policy types during the conversation, saving customers time. It's connected to our API through a function call, so it can give users real-time quotes within the chat interface. All important quote information gets logged into our CRM. (It's publicly available in the U.S. and Canada.) This automation can save individual brokers several minutes before their call with a user. It's currently model agnostic — and we have it running with Gemini 1.5 pro and GPT 4o. We can switch between the two based on performance and price. 

We also have another AI integrated B2B SaaS tool for insurance brokerages. This trust and underwriting automation software, Policy Pilot, has dramatically cut down the time it takes for an underwriter to review an insurance policy — from 30 minutes to about 3 minutes. It uses a vision model to turn complex PDFs into JSON files and uses large language models (LLM) to interpret programmatic checks on the data from the policies. Our time in the Google for Startups Accelerator, particularly the hands-on help we received from our mentor, was a big springboard for the development of this tool. 

Google for Startups: What strategic projects/goals are you using your newly freed-up time to focus on instead? 

Mirzaian: With the time that AI automation has saved us, we can focus on strategic projects such as expanding our market reach, enhancing the customer experience, and developing new product features. We are also dedicating more resources to research and development to stay ahead of industry trends as we continuously improve our AI capabilities.

Google for Startups: How have you used AI to scale faster? 

Mirzaian: We have leveraged AI to scale faster with our new tool, Policy Pilot. By using AI, we've reduced the time that junior developers would need to spend on configuring the tool for provinces outside of Ontario to use. The AI helps amend our code for different provinces, handling small but significant changes efficiently. Vision models also give us a good starting point by parsing out new PDFs from different jurisdictions into JSON. Additionally, our life insurance bot assists in the consumer intake process by answering simple questions about policies, reducing the number of questions brokers need to address, which means they’re now able to take on more customers during business hours.

Google for Startups: How have you used AI to reduce your burn rate? 

Mirzaian: Our Policy Pilot underwriting automation tool has indirectly reduced our burn rate by bringing in direct revenue. This tool has become responsible for 25% of our total revenue 6 months into its soft launch.

Google for Startups: What advice would you give other founders who have yet to adopt AI?

Mirzaian: Adopting AI comes with challenges, particularly when it comes to running LLMs in production-ready applications. It's often more effective to use AI to augment rather than replace programmatic code. But the landscape will change as infrastructure projects expand, models become faster, and inference on chips improves. However, it's essential to avoid overreaching too soon — that would be akin to building a video streaming platform during the dial-up internet era.

Google for Startups: How do you see AI transforming your industry in the next 5 years? What are the advantages of early AI adoption?

Mirzaian: In the next 5 years, advancements in text-to-speech inference will likely lead to more seamless interactions — where customers can converse and receive quotes effortlessly. As LLMs become more accurate and address hallucination issues, they will increasingly automate complex underwriting tasks. 

Improved vision models that can quickly and accurately convert PDFs into JSON will also help lift the insurance industry out of the reliance on PDFs in emails, and provide a bridge for companies that are slow to make their APIs publicly available. This will be similar to how physical scanners helped lift insurance brokerages out of the paper and fax machine era and into PDFs and emails in the 2000s. Early AI adoption will ultimately help companies stay ahead of these trends, gain a competitive advantage, and position themselves as industry leaders.