How Augmented Intelligence’s Symbolic AI Transforms Chatbots

A new alternative to the neural network architectures that power models like OpenAI’s GPT-4 is gaining traction. Known as symbolic AI, this approach uses task-specific rules, such as rewriting text, to address broader challenges. While symbolic AI has historically struggled with efficiency, recent advancements in scalability have made it more viable, allowing startups like Augmented Intelligence to enter the scene.

Symbolic AI’s potential has attracted significant interest, with companies like Orby and TekTonic focusing on enterprise automation and others like Symbolica and Unlikely AI pushing the boundaries of conversational AI. Augmented Intelligence, which recently emerged from stealth mode, is among the latest players in this space, securing $44 million in funding from investors, including former IBM President Jim Whitehurst.

Augmented Intelligence’s Approach to Conversational AI

Augmented Intelligence is developing conversational AI that it claims is more predictable and capable than traditional neural network-based systems. Unlike many existing chatbots that only answer questions, its AI can take actions, such as booking flights or completing tasks on behalf of users.

CEO Ohad Elhelo explains that while platforms like ChatGPT can perform similar actions, they require more manual setup and integration. “There’s a big difference between chatbots like ChatGPT, whose primary goal is to chat with the user, and conversational agents that take actions or work on behalf of companies,” Elhelo told TechCrunch.

The Journey from Delegate to Augmented Intelligence

Augmented Intelligence’s origins date back to 2017, when Elhelo and Ori Cohen co-founded the company under the name Delegate. Initially, Delegate offered an AI-powered app that allowed customers to outsource tasks to low-paid gig workers. The controversial platform received heavy criticism, with workers reporting issues such as poor pay and system glitches, eventually leading the company to pivot toward AI solutions.

With the shift to Augmented Intelligence, the company now focuses on building AI that integrates with a business’s existing tools and workflows to deliver more actionable results. For instance, its AI can retrieve information from external sources and complete tasks without relying solely on its training data, a problem that often limits neural network-driven models.

Privacy Concerns and Data Sources

While Augmented Intelligence claims its AI is more efficient, it has faced questions about data sourcing. The company’s AI was trained on data from tens of thousands of human customer service agents, but Elhelo declined to confirm whether this data came from Delegate or whether those agents were informed or compensated.

Nevertheless, Augmented Intelligence appeals to businesses with strict privacy concerns. Elhelo emphasizes that the AI does not train on a company’s data and only accesses the specific resources it is authorized to use. “It takes into consideration the deploying company’s rule-based instructions,” he said.

This feature could be especially attractive to companies wary of exposing confidential information to third-party AI systems, like Apple, which reportedly banned employees from using OpenAI tools due to concerns about data leaks.

Explainability and Actionability

Another advantage, according to Elhelo, is the explainability of Augmented Intelligence’s AI. The system provides a detailed log of its actions, explaining how and why it responded to queries. This transparency allows companies to fine-tune the AI’s performance and ensure it aligns with their specific needs. Elhelo also asserts that the AI can “eliminate hallucinations,” though experts suggest this claim may be overstated.

Despite the lofty claims, Augmented Intelligence has garnered attention and success. The company recently partnered with Google Cloud, positioning its models on the platform to attract a broader range of enterprise customers.

A Promising Future

Although Elhelo hasn’t disclosed revenue figures, the company recently completed a $10 million fundraising round led by New Era Capital Partners, valuing it at $350 million. While this is a relatively high valuation for a company that has only recently launched its product, Augmented Intelligence’s unique approach to symbolic AI has struck a chord in the crowded AI marketplace.

According to Elhelo, “Traditional models excel at pattern recognition and language generation, but they fall short in situations where the model needs to perform actions, make decisions, or interact with tools. Apollo’s neuro-symbolic architecture and possibilities it unlocks for companies solves those problems.”