Notch’s AI operating system has been accepted into the Microsoft for Startups Pegasus Program. This partnership strengthens how we scale and deploy Notch for enterprises that already run on Microsoft Azure, bringing faster time-to-value, simpler procurement, and enterprise-grade reliability for autonomous resolution.
Notch built for complex, highly regulated environments and was selected based on demonstrated traction and the ability to scale in critical enterprise automation use cases. To date, Notch has processed more than 20 million support tickets across high-volume digital businesses. Within 12 months of deployment, Notch has achieved enterprise-grade end-to-end resolution of up to 87% of tickets without human intervention.
“Our mission is simple: every customer issue fully resolved, every time”
said Rafael Broshi, CEO of Notch. “Joining the Microsoft for Startups Pegasus Program gives us access to Microsoft’s scale and trusted infrastructure, helping us serve more enterprises and accelerate our vision of making resolution, not deflection, the new standard in customer experience.”
What this means for enterprises already on Azure
If your organization has standardized on Azure, the impact of this partnership is straightforward: adopting autonomous customer support becomes easier to deploy, easier to govern, and easier to operationalize at scale.
- Faster deployment in your existing cloud environment: Notch runs on Azure, aligning with the architecture, security posture, and operational controls your teams already use.
- Global performance and reliability: Azure’s global footprint supports low-latency, highly available deployments for enterprises operating across regions and time zones.
- Enterprise procurement paths: With go-to-market through Azure Marketplace, enterprises can discover, procure, and deploy Notch in a way that fits existing purchasing and IT workflows.
- Production-safe iteration: Azure supports the infrastructure foundation we need to safely ship improvements without disrupting live outcomes, which matters when resolution quality is a KPI, not a nice-to-have.
From building insurance for digital assets to productizing autonomous resolution
Explore Notch’s platform on Azure Marketplace.
Notch started in insurance, and not by accident.
Our founding team built the first insurance product designed specifically for digital assets. As we scaled, we needed a way to serve policyholders with speed, precision, and strict compliance. We built internal systems to handle complex service workflows, automate operational decisions, and maintain the auditability and controls required in regulated contexts.
That internal capability became mission-critical. Over time, it became clear it was not just an internal advantage – it was a product.
Notch is the result: governed AI agents that don’t merely respond to customers, but complete actions and resolve issues inside real enterprise systems.
“From day one, we designed Notch as a system that doesn’t just answer but actually acts”
said Yuval Peled, CTO of Notch. “By combining LLMs with deterministic control, our agents can resolve complex customer issues end-to-end. With Microsoft Azure, we can scale that architecture globally with enterprise-grade reliability.”
Built for complex and highly regulated industries
Notch is designed for environments where the cost of error is high, processes are nuanced, and compliance requirements are non-negotiable.
While insurance is where we started and continue to go deep, the platform is built for similarly complex industries like finance and banking, where enterprises need:
- Deterministic process control alongside AI flexibility
- Strong governance, auditing, and escalation paths
- Safe execution of real actions like account updates, refunds, and policy or plan changes
- Consistency across channels including email, messaging, voice, and social support
Notch’s architecture blends advanced LLM capabilities with deterministic rules so enterprises can move beyond deflection and toward true autonomous resolution.
Trust by design: five layers of guardrails and escalation
In regulated and mission-critical operations, trust is not a feature – it is the product. Notch earns trust through a multi-layer guardrail framework designed to keep autonomous resolution safe, compliant, and operationally reliable.
- Layer A: Technical guardrails
Built-in defenses against prompt injection, instruction smuggling, tool abuse, and attempts to extract internal logic. - Layer B: Deterministic access limits
Hard rules determine what a user can view or initiate based on verified state (authentication, role, channel, region, and other identity signals). - Layer C: Deterministic geo and jurisdiction limits
Jurisdiction-aware rules ensure workflows, disclosures, and allowable actions match local regulatory requirements. - Layer D: LLM-as-judge guardrails
Models continuously monitor conversations for boundary risks (loops, frustration signals, disallowed topics) and enforce tenant-specific compliance rules. When risk or uncertainty is detected, Notch escalates. - Layer E: Business limits
Hard stops on execution even for eligible users, including thresholds, rolling counters, and approval requirements for high-risk actions.
Escalation remains a core part of governed autonomy. Notch escalates when knowledge is missing, policy is ambiguous, or risk is elevated, ensuring enterprises get the upside of autonomy without losing control.
%20(1).png)
Scaling globally on Azure and going to market together
As Notch expands into more regulated enterprises and higher ticket volumes, Azure provides the infrastructure foundation to meet stringent performance, reliability, and security requirements.
Notch runs its models and agent orchestration on Microsoft Azure, leveraging Azure’s global scale to support low-latency, high-availability deployments while meeting enterprise expectations around resilience and operational control. In parallel, Notch and Microsoft are partnering go-to-market through Azure Marketplace, making it easier for enterprises to adopt Notch within existing Microsoft environments.
“When you build for insurance first, you learn quickly that reliability is not optional. Customers expect correctness, traceability, and consistent policy enforcement across every interaction,” said Elool Jacoby, CPO and co-founder of Notch. “We built Notch to meet that standard from the beginning, and we’ve carried those lessons into every industry we serve. Partnering with Microsoft helps us scale that promise globally: governed AI agents that resolve complex issues end-to-end, while enterprises retain the controls they need.”
Microsoft’s support through the Pegasus Program and our Azure-first deployment model helps more enterprises adopt autonomous, governed customer support with the reliability, security, and compliance controls required for production at scale. With Notch’s guardrail-first architecture on top of Azure’s global infrastructure, enterprises can move from AI experimentation to real, measurable resolution outcomes.
Why this launch goes beyond the listing
Yes, the listing is live. And yes, it makes procurement easier.
But what we’re most excited about is what sits behind it: the work we’ve been doing with Microsoft’s teams to help make autonomous AI agents not only powerful, but deployable in real enterprise environments where governance, reliability, and control are non-negotiable.
Notch is already automating real customer conversations across chat, messaging, and voice, and those interactions are driving measurable outcomes for companies operating at scale.
If you’re a Microsoft customer and you’re looking to turn your Azure investments into business results – faster resolution, lower operational burden, and a safer path to autonomy – we’re ready to help or you can explore Notch’s platform on Azure Marketplace.
