The xpander.ai agent graph system makes AI agents 4 times more reliable.

Israel’s beginnings xpander.ai has launched its Agent Graph System (AGS), which it says is a key new approach to building more reliable and efficient multi-step AI agents. It is based on basic AI models such as OpenAI’s GPT-4o series.

The goal is to redefine how AI agents interact with APIs and other tools, making it easier for organizations across industries. Access to more advanced automation tasks

Solving the multi-step AI agent challenge

Function call It is the core of most AI agent workflows, allowing models to interact with external systems to perform tasks such as real-time data extraction or execution.

However, these interactions often falter in the face of complex API schemas or unpredictable responses. This leads to inefficiency and errors.

xpander.ai’s Agent Graph System provides a structured solution to these challenges using a graph-based workflow that guides agents through the appropriate API calls step-by-step.

Instead of presenting all the tools available at every step, AGS intelligently limits the options to those that are relevant to the current context of the work. This greatly reduces out-of-order or conflicting function calls.

Ran Sheinberg, Co-Founder and Chief Product Officer of xpander.ai, explained in an interview with VentureBeat, “With AGS, we ensure that agents only use the relevant tools at each stage. and follow the correct schema. It enforces precision and efficiency.”

Sheinberg previously worked at other startups. Several locations and leads the core solution architecture at Amazon Web Services (AWS), leading large-scale computing projects with enterprise customers.

Democratizing AI Agents

xpander.ai aims to make agent-based AI development accessible to a broad audience. “We aim to create an accessible platform that allows anyone to create AI agents, experiment with technology, and automate repetitive tasks to focus on what truly matters,” David Twizer, co-founder and CEO of xpander.ai, said in the same interview.

The company also offers AI-ready connectors that easily integrate with NVIDIA NIM (Nvidia Inference Microservices) and other systems. These connectors complete the API tool with detailed documentation. Operational code and schema reduce the technical burden on developers while improving runtime accuracy.

“Once the setup is complete. You can connect to an AI system that supports calling functions,” Twizer said. “It is important for us to design technology that meets our customers’ needs. and offers the flexibility to upgrade models over time.”

Twizer previously worked at AWS where he was a Principal Solutions Architect and led the go-to-market innovative AI sales architecture.

Key Benefits and Real World Impact

In benchmarking tests, xpander.ai showed that AGS,matched to a representative interface. It helps AI agents achieve a 98% success rate in multi-step tasks. This compares to just 24% for agents using traditional methods.

These agents completed their workflows 38% faster and used 31.5% fewer tokens, reinforcing AGS’s ability to reduce costs and improve efficiency.

A real-world example of an AGS application involves a benchmarking task where an AI agent must research different companies. across platforms like LinkedIn and Crunchbase, then organize the results in Notion AGS, streamlining the process. It ensures that tools are used in the correct order and plans are followed consistently.

“We provide a complete AI agent that can create interfaces for any system,” Twizer added. “For the first time, data interfaces are AI native, which addresses critical problems that The world is struggling too.”

AGS’s role in AI agents

xpander.ai positions AGS as a key step in the evolution of agent-based AI, enabling tools like Nvidia NIM microservices to integrate more seamlessly with enterprise systems.

“AI agents will need APIs for synchronous use cases involving complex data structures for which traditional UIs are inadequate,” Sheinberg said.

xpander.ai transforms the way AI agents handle error handling and context persistence through AGS. By embedding fallback options directly within the graph structure, AGS enables agents to retry failed operations or transition into Alternative workflow without human intervention by maintaining job stability

This level of reliability ensures that AGS-equipped agents not only respond, but also respond. but can also be changed and can handle even the most unpredictable workflows.

Building the future of AI workflows

xpander.ai’s AGS release alongside agent-based interface It represents a major leap forward for multi-step AI agents.

By enabling structured and adaptive workflows. and improving complex API interactions, AGS sets a new standard for reliability and performance in automation.

As the company continues to grow Various tools The company also promises to enhance the potential of various businesses. To harness the full potential of AI-powered workflows