JP Vertil

We've been deep in the grind, collaborating with exceptional partners to bring you a Gaya that's faster, sharper, more powerful, and, dare we say, more captivating. Our latest update is a testament to real-world challenges and invaluable feedback—delivering automation that's as efficient as it is elegant. A heartfelt thank you to our early adopters and agency partners for inspiring us to elevate our game! You know who you are :) Discover the enhancements in this short video - longer version on our website The next major product update is what you've all been waiting for: the launch of Gaya Commercial Lines. The beta is going great, and we can't wait to share it with you. #InsurTech #automation #builtwithpartners #Insurance #AI

Running a remote team has been one of the best decisions we’ve made at @getgaya, and one of the hardest to get right. One big non-negotiable for us: at least 3 hours of overlap every day across the entire team. In practice, it’s often more. That shared window matters a lot. It gives us enough time to unblock each other, make decisions quickly, and actually feel like we’re building together instead of just passing work across time zones. The upside is obvious: you get access to incredible talent you’d never find if you stayed local. Our team stretches across Romania, the US, Lebanon, Haiti, Ukraine, the Philippines, and Tanzania. That’s pretty special. The harder part is what remote work naturally takes away. You miss the random conversations. The quick check-ins. The little moments that build trust without anyone trying too hard. We’ve tried to be intentional about that. Outside of deep work, we have a pretty huddle-friendly culture on Slack. Quick calls are encouraged when something is easier to talk through live. We also have weekly team meetings that are less about status updates and more about staying connected, celebrate wins and each other. It’s not perfect. I still think remote teams have to work harder to create the kind of cohesion that can happen more naturally in person. But done right, I think the tradeoff is worth it. Curious how others handle this: What are your best practices for building strong culture and fast execution on a remote team?

Introducing the Tasklet sandbox VM. Every agent now gets its own Linux environment with Python, ffmpeg, imagemagick, and a persistent filesystem. Create, process & transfer files, use code to analyze data, make network requests, install packages & more.

See @getgaya in action at ITC Vegas 2025! We’re bringing live demos, and a fun copy and paste competition for some very nice prizes! If you’re exploring how to quote faster (without adding complexity), stop by our booth, or better yet, book a 1:1 early to experience it firsthand. Join us and see why insurance brokers are choosing Gaya to quote faster. #ITCVegas #Insurtech #GayaAI #Automation

Today we're launching Tasklet — an AI agent for automating your business. Unlike ChatGPT, @TaskletAI actually does the work for you: connecting to your tools, triggering automatically, and handling tasks while you sleep.

@AndrewYNg @guardrails_ai @ShreyaR I’d love to see a module on comparing probabilistic vs. deterministic guardrails. i.e. what’s possible when misbehavior is impossible by design vs. just filtered after it happens

New short course: Safe and Reliable AI via Guardrails! Learn to create production-ready, reliable LLM applications with guardrails in this new course, built in collaboration with @guardrails_ai and taught by its CEO and co-founder, @ShreyaR. I see many companies worry about the reliability of LLM-based systems -- will they hallucinate a catastrophically bad response? -- which slows down investing in building them and transitioning prototypes to deployment. That LLMs generate probabilistic outputs has made them particularly hard to deploy in highly regulated industries or in safety-critical environments. Fortunately, there are good guardrail tools that give a significant new layer of control and reliability/safety. They act as a protective framework that can prevent your application from revealing incorrect, irrelevant, or confidential information, and they are an important part of what it takes to actually get prototypes to deployment. This course will walk you through common failure modes of LLM-powered applications (like hallucinations or revealing personally identifiable information). It will show you how to build guardrails from scratch to mitigate them. You’ll also learn how to access a variety of pre-built guardrails on the GuardrailsAI hub that are ready to integrate into your projects. You'll implement these guardrails in the context of a RAG-powered customer service chatbot for a small pizzeria. Specifically, you'll: - Explore common failure modes like hallucinations, going off-topic, revealing sensitive information, or responses that can harm the pizzeria's reputation. - Learn to mitigate these failure modes with input and output guards that check inputs and/or outputs - Create a guardrail to prevent the chatbot from discussing sensitive topics, such as a confidential project at the pizza shop - Detect hallucinations by ensuring responses are grounded in trusted documents - Add a Personal Identifiable Information (PII) guardrail to detect and redact sensitive information in user prompts and in LLM outputs - Set up a guardrail to limit the chatbot’s responses to topics relevant to the pizza shop, keeping interactions on-topic - Configure a guardrail that prevents your chatbot from mentioning any competitors using a name detection pipeline consisting of conditional logic that routes to an exact match or a threshold check with named entity recognition Guardrails are an important part of the practical building and deployment of LLM-based applications today. This course will show you how to make your applications more reliable and more ready for real-world deployment. Please sign up here: deeplearning.ai/short-courses/…

@SRajdev Hi Swapan, I’ve been down the same rabbit hole :) One strand I keep circling back to: what if guardrails weren’t just reactive (prompts + filters), but deterministic by construction? Happy to share what I’ve sketched out (and learn what holes people see).

I went down a bit of a rabbit hole recently trying to understand AI guardrails. What started as “how do I stop bad outputs?” turned into a deep dive into architecture, safety frameworks, unintended behaviors, and a dozen real-world incidents where things almost went off the rails. Most people treat guardrails like seatbelts: a safety add-on. But in agent systems that plan, reason, call tools, and store memory, guardrails have to be part of the architecture, not just wrappers around LLM outputs. I pulled everything I was learning into a blog post and realized that before talking about how guardrails work, we need to understand what exactly we're trying to protect. 🧩 Here are the 6 key attributes that good guardrails should defend: - Accuracy: prevent hallucinations & disinformation - Privacy: avoid sensitive data leakage - Security: block injections, unsafe code/tool access - Safety: prevent harmful or dangerous behavior - Compliance: align with legal/regulatory standards - Fairness: reduce bias in reasoning and responses Read the full post to explore how guardrails work and why thinking of them as runtime actions across your entire agent stack is a much more powerful model than just filtering input and output. Link to the full post here: srajdev.com/p/multi-layere…

Did you know? AIG is building a digital twin of its entire business with GenAI. Early tests cut some claim processing times from days to hours. Glad to see more AI push in insurance. #Insurance #AI insurancejournal.com/news/national/…

Glad to see @getgaya driving automation in large brokerages and making real change. Thank you @BrokerTechVen for the feature. Back to perfecting our supercopy and superpaste product and scaling it to all lines. Stay tuned for the many lines we have been quietly launching!

4/4 You can't ignore AI if you want to evolve. But you have to be intentional and set the right expectation. Huge thanks to the Haitian American Chamber of Commerce of Florida @HACCOFL and Tribaja for the invitation!

3/4 1. To get started with AI personally: Be curious and try out different tools. 2. To get started with AI in your business: First, establish your baseline processes. AI will amplify your systems, but it won't fix them. 3. To identify opportunities: Ask yourself what is mundane/repetitive and seek tools that can solve key pain points.

I often here the same questions about AI: “Where do I start?” “Which tools actually matter?” “How do I apply this to my business?” The truth is, AI feels exciting but overwhelming.

Founders need a 9 to 5 too. Lately I’ve been spending my days at a coworking space. It helps me timebox and knock out focused work without distractions. Then I head home… and start shift #2. The change of environment gives me just enough of a reset to keep going.

Back at Notre Dame for the first time since graduating. Great to see faculty who’ve always supported me and to watch the @ndideacenter's Race to Revenue accelerator wrap up. ND is taking big steps in entrepreneurship! Go Irish! 🍀

A good read on AI agents for full-blown automations. Having clear boundaries and keeping humans in the loop for critical workflows remains essential. utkarshkanwat.com/writing/bettin…

"Big opportunities in life have to be seized. We don't do very many things. But when we get the chance to do something that's right and big, we've got to do it. And even to do it on a small scale is just as big a mistake, almost as not doing it at all" - Warren Buffett

At @EqualVentures ' #InsuranceCapitalSummit last week. Insurers crave AI automation but as pointed by @sychou , they demand observability + control. At @getgaya we ship AI only when we can explain, predict & trace every step: magic for users, never a black box for us. Thanks @achadroff & team!

good stuff @obed_lamy !

Over 250 tech CEOs (from Microsoft to Doordash) are calling on states to make CS & AI part of high school graduation requirements. Good push. Source: ajsai.substack.com

Most product challenges aren’t technical. They’re human. A friend recently asked about the hardest part of building @getgaya . My answer is in making it feel simple—even when it’s powering complex workflows on clunky, legacy insurance portals. We could’ve layered on UI for every edge case. Instead, we chose restraint. Less “if this, then show that.” More “just works.” That’s why our tagline is: Simple. Yet Powerful.