The Compliance Engine: How Saphira AI Eliminates the Safety Blockade for Autonomous Systems

Blog
Written By Greg Toroosian

The largest hurdle for robotics and autonomous systems today is not a technical one; it is a question of trust, risk, and compliance. Many innovative systems remain sidelined because the process of achieving safety certification is slow, expensive, and non-scalable. This critical bottleneck is what truly hinders widespread technology adoption, preventing valuable innovations from reaching the market and generating economic returns.

On Episode 101 of The Machine Minds Show, host Greg Toroosian, founder of Samson Rose, discussed with Akshay Chalana, CEO and Co-founder of Saphira AI. They talked about how the company is using Artificial Intelligence (AI) and agents to automate compliance, dramatically reducing the time and cost required to bring safe, regulated hardware products to market.

In this conversation, Greg Toroosian and Akshay Chalana explored why safety and certification are the hidden blockers for robotics at scale; how Saphira’s agent-based platform turns Computer-Aided Design (CAD) files, schematics, and risk regulations into an actionable compliance plan; real-world wins (industrial arms, home humanoids) and what “shift-left safety” looks like in practice; and the fascinating career twists from Taekwondo black-belt to hedge-fund engineer to startup founder.

From ML Compliance to Hard Tech: Akshay Chalana's Path to Safety

The journey that led Akshay Chalana to found Saphira AI is a masterclass in diving into the complex technical and regulatory domains that few others wish to tackle. His career experiences provided a deep, cross-functional understanding of system integrity and failure points, ultimately defining the scalable compliance problem Saphira AI solves today.

Akshay’s professional path began in software engineering, with time spent in Machine Learning (ML) at Meta and Apple. Critically, he focused on the compliance side of ML, which included rigorous data set management for complex products like Vision Pro and Siri. This early exposure to the fragility of cutting-edge systems was followed by reliability engineering at Tesla, which directly inspired the company's direction, and a pivotal role in New York, building a compliance engineering team at the hedge fund Citadel. These cumulative experiences introduced him deeply to the problem of data-driven robustness and safety of complex systems.

Akshay discovered that a core issue in engineering is that building things without a key stakeholder or a clear use case often leads to burnout. It took a long time to discover a problem that was clearly unsolved and a demonstrable need that people were struggling with. That discovery, the struggle with archaic requirements management systems like IBM DOORS, which are manually intensive, slow, and non-integrated, became the founding spark for Saphira AI. 

By solving this bottleneck, Akshay sought to accelerate the pace of hardware innovation. On a personal note, Akshay holds a black belt in Taekwondo and is an avid mountaineer, having climbed Kilimanjaro, a background that underscores the discipline and systematic approach necessary for tackling complex compliance challenges.

Saphira AI: The Compliance Engine for Robotics

Saphira AI is designed to act as a compliance navigator for the robotics industry. The analogy captures the mission of Saphira to demystify and simplify a process that has historically required expensive, specialized consulting firms. It automates the path for any stakeholder, from a robot builder or integrator to an industrial facility user, who has specific compliance obligations.

Akshay notes that historically, only highly regulated industries, such as automotive and pharmaceutical manufacturing, have been compelled to care deeply about compliance. 

Today, however, almost everyone is reaching production, deploying robots into uncontrolled industrial facilities, and facing real-world obligations like Occupational Safety and Health Administration (OSHA) and insurance mandates. 

This massive shift in deployment requires a scalable compliance solution that simply did not exist before.

The Saphira AI Advantage Over Tradition

Saphira AI is fundamentally a software platform, not a consulting service. This distinction is crucial: by packaging the entire procedure into a dynamic tool, it allows compliance efforts to move much more quickly and be cheaper than traditional, human-intensive alternatives. This technological leverage means Saphira can help products reach final certification in a fraction of the time required by manual processes.

Importantly, Saphira AI provides transparency and empowerment. While many organizations hire expensive internal safety experts or external consultants, Saphira AI provides the tools to build procedures and teach clients how to maintain compliance themselves. This approach enables clients to build up their internal processes and design safer products more quickly, moving beyond dependency on external advisors and building lasting internal competency.

Technology Deep Dive: AI Agents and Structured Reasoning

The core of Saphira AI’s technology is the use of specialized AI Agents, which are structured reasoning systems. Akshay emphasizes that while Large Language Models (LLMs) are useful for general information, they are not trustworthy for compliance work, often missing critical details or introducing subtle inaccuracies. Compliance requires verifiability and precision, which LLMs alone cannot guarantee.

Saphira AI’s agents operate by leveraging LLMs for small, defined components but utilize a range of specialized tools for complex computational processing and structured data analysis. 

The system takes unstructured data, such as design documents, CAD Models, schematics, and videos of prototypes, and outputs a clear description of which standards are applicable.

CAD, which stands for Computer-Aided Design, is the use of computer systems to assist in the creation, modification, analysis, or optimization of a design, typically producing 2D drawings and 3D models used in engineering and manufacturing.

The process includes building a tool that:

  • Interprets CAD: It interprets CAD in a usable format for a Vision-Language Model (VLM), allowing the VLM to interface with itself and understand the robot's physical structure.

  • Redirects Focus: As the system analyzes the CAD, it can redirect the angle or zoom in on specific components or gaps for analysis, ensuring no safety-critical part is overlooked.

  • Structured Reasoning: All this information is brought into a structured reasoning exercise, which is key to producing a reliable, actionable compliance plan that auditors can trust.

The end goal is to create the "safety engineer that is going to be in the loop" when agents are developing and iterating on the whole system, ensuring safety is prioritized by the AI from the beginning of the design phase.

Real-World Compliance Success and The Shift Left Imperative

Saphira AI has delivered significant compliance outcomes in the industry, including assisting Robco, the modular industrial arm manufacturer, in preparing for the ISO 10218 certification, which is the key standard for industrial robots. They also collaborated with the humanoid company 1X to build the safety concept for NEO, the home-deployed humanoid robot.

In both scenarios, Saphira AI begins by mapping the complete landscape of relevant standards. It then defines the safety architecture, which includes components such as e-stops and protective zones. In addition, the system generates detailed risk assessments, specifically using standards like ISO 12100 or ANSI B11. This comprehensive assessment is then used to define and validate the necessary safety functions. The entire process results in the final technical report required for regulatory compliance.

Akshay believes Saphira AI will fundamentally influence the hardware industry by tackling the core problem of technology adoption: risk.

  • Unlocking Deployments: The goal is to flip the compliance problem from a pure design hurdle to a dynamic one. By monitoring and regulating safety through deployment in the field, it becomes possible to hold systems accountable and unlock massive deployments that were previously deemed too risky to scale.

  • Engineering Outcome: Shift Left: The most important vision is to "shift left" the compliance problem. This concept, borrowed from the software security world, means every safety consideration must be easily integrated at the very beginning of the development life cycle, specifically at the time of CAD design, software design, and simulation. This is only possible if the interpretations are readily available when the engineering work is happening. This is critical because if you write code and you move really fast, you must have tools that ensure compliance does not slow down innovation. If compliance checks are left until the end of the process, fixing issues becomes exponentially more costly and time-consuming, grinding development velocity to a halt.

Talent, Culture, and the Road Ahead

Akshay is particularly excited about the trend of moving safety from an isolated system to a more integrated, modular architecture. Isolated systems previously relied on separate compute and sensors.

He notes that new platforms, such as the Nvidia IGX Orin, exemplify this shift. This deep integration allows a single system to architect its own safety software and perception stack. By running everything in a centralized system, compliance becomes easier to manage and enforce across the entire fleet.

Obstacles and The Future of Risk

On the other hand, Akshay sees numerous obstacles in achieving widespread AI safety compliance, primarily due to the lack of clear frameworks for how to properly mitigate the risks produced from these AI analyses. The industry needs to define how the objective, verifiable outcomes of AI risk analysis can be standardized and leveraged to:

  • Price Insurance: How can the output of an AI risk analysis influence the pricing of insurance for these systems? Standardized, objective risk data is essential for insurers to calculate premiums fairly and accurately.

  • Attribute Liability: How can this be used to properly attribute liability in the event of an incident? Clear, documented compliance data is the bedrock for establishing legal responsibility when a complex, autonomous system fails.

  • Influence User Behavior: How can the information reduce the chance of things going wrong by providing the right data at the right time? Real-time safety insights need to be fed back to operators to minimize human error and promote safe operational practices.

The core challenge remains ensuring that all the right stakeholders, from technical to legal, are at the table so a comprehensive story is told that proves these products are safe.

Hiring for Urgency and Risk

When building his team, Akshay focuses on two primary characteristics:

  1. Risk-Taking: A mindset where a person is willing to "stake your name on an opinion" that is aspirational, and then take the initiative to figure it out and achieve it. This requires internal courage and accountability.

  2. Urgency: This is the most valuable asset a startup can provide. It is the belief that one owns everything produced or recommended and feels the accountability to fix things, no matter the time of day. Akshay notes that this specific type of urgency comes from having agency experience, a viewpoint Greg Toroosian concurred with. Unlike an external consultant, who might deliver a report and step away, someone with an owner's mindset has inherent personal accountability. They understand that their name is staked on the outcome, and they feel the imperative to fix problems immediately, regardless of the time or internal obstacles.

Akshay shares that every person on his six-person team has been a founder or has run their own business, highlighting that having agency experience is critical to having immense pride in what one is putting out in the world.

For aspiring founders, Akshay’s final advice is simple. He emphasizes that to succeed, you must embrace core principles supported by clear, actionable steps:

  • Embrace risk-taking and putting yourself on the line, as nothing comes without it

  • Move as fast as possible.

  • Close the loop on every project.

  • Embrace the worst-case scenarios with a clear plan and framework.

Notes from the Podcast Episode: Saphira AI in a Nutshell

  • Saphira AI's core purpose is to remove the biggest hurdle blocking new robotic products from launching: the slow, expensive, and manual process of getting safety certified by regulatory bodies, like getting certified for ISO standards or meeting OSHA requirements.

  • Instead of hiring a human safety consultant for months, Saphira uses AI agents to speed up this process:

    • Input Analysis: The system takes all the robot's design files, such as CAD models and schematics, and all the relevant safety regulations.

    • Compliance Plan Generation: The AI analyzes the design against thousands of rules to automatically generate a comprehensive risk assessment and compliance plan. It identifies exactly where the design is non-compliant and what needs to be fixed.

    • "Shift-Left" Safety: By checking the design in the very beginning or shifting the compliance work to the left of the development timeline, they help companies build products that are safe from the start, saving massive amounts of time and money that would otherwise be spent fixing costly safety issues at the end.

  • In short, Saphira AI provides the software tools to help robotics companies get their hardware products to market faster and cheaper while ensuring they meet all necessary safety standards.

Explore the Innovation Driving AI-Powered Compliance

By embracing the shift-left philosophy and AI-powered tools championed by leaders like Akshay, technology professionals and companies alike can successfully navigate the evolving landscape of intelligent automation and maximize their impact.

Stay Informed with AI and Robotics

  • Gain deeper insights by subscribing to The Machine Minds Show on YouTube. Start listening today to access visionary conversations that are shaping the future of technology and leadership.

  • Make sure to also follow on Instagram and Facebook to track valuable market insights and updates on upcoming episodes.

  • Connect with Samson Rose founder and The Machine Minds host, Greg Toroosian, to stay abreast of the latest industry trends and discussions.

Take Action Now: Secure Your Role in the Hard Tech Revolution

If your expertise lies at the cutting edge of Robotics, AI, or Hard Tech, your next defining leadership role awaits.

Samson Rose is the specialized talent search partner for high-velocity innovation. We meticulously connect exceptional leaders from Directors to C-Suite with the visionary companies defining the future of autonomous systems.

Stop searching, start building.

Greg Toroosian
Next

Robots for People: How Robust AI Redefines Human-Centered Automation