The future of agentic AI in business and the path to industrialization
In this exclusive interview, Paul Sinai, CEO of Blaxel.ai, shares his vision of making his company “the AWS of agentic AI” in a future where every business process is driven by autonomous agents.
How do you see the world of intelligent agents evolving within companies ?
We are experiencing a technological transformation comparable to the rise of the Internet : soon, every company and every business process will be powered by autonomous agents.
Agents will not simply execute tasks, they will code, deploy, and orchestrate other agents. This creates an exponential dynamic. We are moving from a world of static websites to one where billions of agents continuously interact with systems.
The next standards of productivity and automation will be built around these technologies.
What do you see as the key use cases for agentic AI ?
The first major use cases involve automating administrative and repetitive tasks : IT access management, user support, invoice reconciliation, expense management, and incoming email processing.
We are already seeing new startups being built with very small teams because they automate accounting, finance, and IT operations from day one.
Agentic AI also enables the creation of “AI First” applications, where interfaces are redesigned from the ground up around intelligence itself.
What are the main technical challenges you identify ?
The challenges are numerous :
- Infrastructure : Traditional cloud environments were not designed to host continuously communicating agents. We need to reinvent the equivalent of a CDN (Content Delivery Network) for agents.
- Performance : Every second saved during agent execution allows more customers to be served. Optimizing latency is therefore critical.
- Hardware : Current GPUs such as the H100 are expensive, degrade quickly, and were not originally optimized for continuous inference workloads. The future may rely on alternative architectures such as TPUs, FPGAs, or RISC-V.
- Security and supervision : Organizations must ensure that agents operate securely by isolating them, for example through containers, while continuously monitoring them to detect deviations or abnormal behaviors.
And what about the business challenges ?
The primary challenge is cultural : companies must learn to trust agents with critical tasks.
There is also a massive need for training and support. Organizations will need to explain these technologies, frame their usage, and build trust around these new tools.
Another major challenge lies in integrating agents with existing systems such as ERP, RPA, and CRM platforms, which were not originally designed to interact with autonomous agents.
What is Blaxel.ai’s value proposition in this context ?
Blaxel.ai was built on a clear observation : current cloud infrastructures are not designed to support billions of autonomous agents.
Our ambition is to become “the AWS of agentic AI” by providing infrastructure specifically optimized to host, manage, and orchestrate intelligent agents at scale.
We offer a complete set of building blocks, including cloud infrastructure, orchestration, monitoring, and security solutions, all specifically designed for these new use cases.
We provide the full infrastructure stack required to make agentic AI viable at scale : hosting optimized for low latency and high agent density, orchestration tools for agents through MCP Servers, advanced supervision capabilities including activity monitoring and redundant access to LLMs, as well as reinforced security through isolated containers capable of safely executing generated code.
Our goal is to enable any company or developer to deploy agents in a simple, reliable, and scalable way.
How do you see the relationship with existing major players such as hyperscalers, ERP vendors, and RPA platforms ?
Hyperscalers such as AWS, Azure, and GCP will remain essential, but their infrastructures were not originally designed for agentic AI.
Over time, they will either need to evolve significantly or collaborate with specialized players like us.
As for ERP and RPA ecosystems, many legacy vendors may struggle to adapt because their platforms were built for a pre-agentic world and tend to remain rigid in their architecture.
The companies that will succeed are those willing to open their platforms to autonomous agents through MCP-compatible APIs.
What advice would you give to companies looking to integrate agents into their processes ?
Start small and move quickly. A typical roadmap could look like this :
- Build an initial no-code agent to automate a simple task, such as publishing a blog post, filling in a CRM, or extracting information from a document.
- Measure the actual productivity impact.
- Gradually move toward custom development if necessary.
- Train your teams to understand and supervise agents, even without coding skills.
Companies should not aim to fully automate an entire department from day one. Instead, they should prioritize rapid iterations on high-volume, low-complexity tasks.
How do you see the broader technology landscape evolving with agentic AI ?
We are about to witness a shift from the “search engine” era to the “answer engine” era : instead of searching for information ourselves, we will delegate this task to agents.
This evolution will have major consequences across entire sectors including digital marketing (SEO and SEA), e-commerce, and customer relationships.
At the same time, technical infrastructures will undergo deep transformation, with new standards emerging around identity management, communication, and supervision, such as MCP.
One final thought ?
The agentic wave will be disruptive.
Companies capable of experimenting quickly, pragmatically integrating agents, and preparing their infrastructures accordingly will survive and thrive. The others risk being overwhelmed.
To learn more, visit the Blaxel.ai website.
To explore these topics further and identify the impacts and opportunities relevant to your organization, contact our expert :
Morand Studer (Managing Partner et Expert AI)
