What Makes C3 Generative AI Unique

As ChatGPT and other LLMs have soared in popularity, so has attention to the fact that its seemingly remarkable answers are sometimes flat-out wrong. The LLMs that interpret a user’s queries don’t have a good way to check the accuracy of source data, nor are they designed to tell you when they don’t know an answer. Similar to the iPhone keyboard’s predictive-text tool, LLMs form coherent statements by stitching together data — words, characters, and numbers — based on the probability of each piece of data succeeding the previously generated piece of data.

When LLMs do not know an answer, they frequently make one up. The results are pure fiction – technically known as hallucination. Any Generative AI solution for the enterprise must prevent hallucination.

Instead of relying on an LLM to generate an answer, the LLM should effectively hand off the query to an underlying orchestration agent that retrieves the answers from deep learning models already applied to an enterprise’s data. C3 AI solves this by not relying on the LLM to provide the answer. With C3 Generative AI, the LLMs have no direct access to the data. The LLM is enabling the enterprise AI software already applied to an organization — and thus provides reliable responses. Even better, an ideal generative AI system for the enterprise should tell a user when it doesn’t know an answer instead of generating an answer strictly because that’s what it’s trained to do. C3 Generative AI only provides an answer when it’s certain the answer is correct. If it can’t find an answer, it tells you so. Deterministic answers. No hallucination.

C3 AI’s Generative AI Journey

C3 AI began working with generative AI long before ChatGPT and the term large language model (LLM) became part of the common lexicon. In 2020, the U.S. Department of Defense’s (DoD’s) Missile Defense Agency (MDA) asked C3 AI to model the trajectory of incoming hypersonic missiles to speed up the MDA’s ability to anticipate incoming missiles. Using traditional supercomputer methods, it could take 12 hours to make a few dozen simulations. The MDA’s goal was to have thousands of trajectories modeled in a matter of minutes.

The MDA’s unique needs required a new approach to deep learning models—an AI method that teaches computers to make predictions and decisions without explicit programming. To run such powerful simulations quickly, and at scale, also required the adoption of transformer models. First described in a Google paper in 2017, transformer models are powerful neural networks that can learn context by tracking relationships in data; they are the foundation of generative AI.

C3 AI’s then-novel use of transformer models for the MDA was an early use of generative AI technologies applied to enterprise-scale (and enterprise-secure) machine learning (ML) use cases. C3 AI achieved the MDA’s request, substantially reducing the time it took to create simulations of hypersonic missiles. The MDA was also able to provide these surrogate models to defense contractors such as Raytheon Technologies and Lockheed Martin so they could better model missile trajectories to build effective defense systems.

The more significant breakthrough for C3 AI came more recently; it also began with a request from the U.S. military. In September 2022, C3 AI CEO Tom Siebel received an email from an official at the DoD.

“I want to be the Google for DoD,” the official wrote. “A user asks a question and gets an answer. How do I make that happen?”

“Good idea,” Siebel replied. “Give me a couple of weeks to ideate this and propose a plan.”

Siebel and his team sketched out a new way for users to communicate with the disparate data sources and machine learning-models unified by the C3 Agentic AI Platform by incorporating the look and experience of a common search interface.

A New Way to Use Large Language Models

The LLMs that underlie generative AI technologies such as ChatGPT, Google Bard, and others, presented C3 Al with a valuable starting point for a broader set of solutions for the enterprise. C3 Generative AI is designed to be LLM agnostic, which is important as LLMs evolve, and new models are introduced—something that is happening at a rapid pace.

Currently, C3 Generative AI can leverage LLMs available through the cloud providers, such as Azure’s OpenAI Service, AWS Bedrock, and Google’s PaLM 2. C3 Generative AI also uses C3 AI’s own fine-tuned LLMs, including FLAN-T5 and Falcon 40B models. With our fine-tuned LLMs, C3 AI can also support highly secure, internet air-gapped environments required by some customers.

C3 AI doesn’t rely on an LLM the way consumer applications do. LLMs such as ChatGPT interact directly with the data sources they’re trained on, such as publicly available data on the internet. With C3 Generative AI, the LLM serves as a kind of AI librarian that figures out and interprets a user’s question and then directs the query to a retrieval agent that works with the C3 AI Type System to figure out where (and how) to get the best answer amid an enterprise’s data corpus.

C3 AI’s unique implementation of generative AI benefits from its proven enterprise AI technology, that the company spent more than 14 years and roughly $1.5 billion developing. C3 AI’s architectural foundation derives from a proprietary, model-driven architecture that eases and accelerates the development and delivery of enterprise AI applications at scale, empowering business decision makers with predictive analytics, resulting in substantial economic benefit.

For generative AI models to be truly transformative for the enterprise, C3 AI combined the LLM with its deep experience integrating, unifying, and modeling enterprise data for machine learning including predictive analytics, supervised learning, unsupervised learning, and deep learning. Enterprise AI, a category C3 AI established, is broadly recognized today as a large and growing market. (IDC projects the AI software market will nearly double between 2022 and 2026, reaching $791 billion.)

C3 AI’s novel use of large language models (LLMs), retrieval agents, and C3 AI Type System or object models, makes C3 Generative AI the only fit-for-business generative AI technology available today. It’s also constantly evolving. C3 Generative AI includes a feedback system, so users can let the system know if a response needs refining based on an updated piece of information, such as data from a newly installed sensor. In addition, unlike ChatGPT, C3 AI isn’t working off a set of data captured at a given moment in time. The data the system interprets are dynamic — as the data changes, so too do the answers provided. Such fluidity is critical for large enterprises, where the underlying information (data) is frequently changing.

While the impetus for C3 to create a generative AI solution started with a request to be internet-search-like, C3 Generative AI for the enterprise is far different. The search-like interface was used for a simple reason: It’s familiar and intuitive. But what the user gets from C3 Generative AI is more than enterprise search.

C3 Generative AI serves as a knowledge assistant, with domain expertise, that can orchestrate APIs, tools, and help users take action.

And it all happens rapidly, with a user’s prompt, or query, flowing down to the learning models already applied to an enterprise’s datasets.

Giving an enterprise user an intuitive way to ask questions of all an enterprise’s data expands the value of enterprise AI technology to far more people across an organization.

What Makes C3 Generative AI Unique

Standard LLMs Vs C3 Generative AI’s solution:

Six critical capabilities differentiate C3 Generative AI from consumer generative AI applications – making it the only generative AI solution that business and their users can rely on to get consistent, trusted, and secure answers to all of their business questions.

1. Consistent Answers

One problem with GPT LLMs is they frequently provide random, inconsistent responses. In fact, two people can ask the same question and get different answers. The LLMs aren’t designed to provide precise, deterministic answers necessary for any commercial or government application.

Generative AI for the enterprise must provide consistent answers and handle nuance. If two people ask the same question, or even phrase it in slightly different ways, the system needs to connect to an overall technical architecture that ensures those queries are routed to the applicable data sources within the enterprise to generate identical and correct answers. C3 Generative AI for the enterprise needs accesses an organization’s entire corpus of data, including ERP, CRM, SCADA, text, PDFs, Excel, PowerPoint, and sensor data.

To produce deterministic, traceable, accurate answers, C3 Generative AI doesn’t rely on the LLM to come up with the answer. Instead, the LLM delivers the query to a different system (a retrieval model connected to a vector store) to understand which documents or data sources are the most relevant. With this method, the answers generated always comes from the best sources in the system at that given moment, ensuring answers that an enterprise user can trust.

2. Full Traceability

The speed at which LLMs provide seemingly polished answers can create the impression those answers are accurate and authoritative. They are not. In addition to sometimes providing random answers, consumer LLM applications lack what’s known as traceability. Put simply, they don’t have the ability to track and provide information about the datasets they’re relying on to generate answers. They are not designed to do so, making it all but impossible to check the veracity of any claim.

It is a level-zero requirement that any generative AI solution for the enterprise always indicates where its answers come from. It should synthesize an answer and provide a chat interface so the user can dive deeper with follow-up questions — and the dashboard needs to provide instant traceability. C3 Generative AI provides all of that. It’s similar to searching the web, where the user can decide whether a website that’s served up is credible. If business users are going to take action based on answers from generative AI, they must be able to trust the software’s response.

3. Full Access Controls

Every business and government agency gives different access to different employees. The general counsel can see documents other employees can’t. The same is true for the CFO and everyone in various functions across an organization.

One of the significant opportunities with generative AI is that people throughout an organization—not just business and data analysts, for instance—can take advantage of the technology so they can do their jobs far better and far more productively. It’s also an opportunity to upskill more junior employees. To make all this possible, a generative AI enterprise solution must provide rigorous access controls to ensure each user gets fast and accurate answers while maintaining security and data confidentiality for your entire enterprise. C3 Generative AI manages access via a comprehensive role-and-policy based framework built into the underlying C3 Agentic AI Platform.

4. Protection From LLM-Caused Leakage of Proprietary Information

In offices everywhere, millions of workers are using ChatGPT and other LLM-based generative AI tools to help them do their jobs, even as their bosses and IT departments are often unaware. What workers likely don’t realize is that any time they engage with a consumer generative AI service, what they feed into the system—the prompts they enter to help generate marketing announcements, say, or create code requests—are then stored on external servers, resulting in IP and trade secret exfiltration.

In May 2023, Samsung banned employees from using ChatGPT after discovering some of its sensitive code was uploaded to the internet. Other companies have also cracked down on the use of ChatGPT and other generative AI tools and the like. Such risks don’t exist with C3 Generative AI. With C3 AI’s solution, the deep learning models that derive answers do so on the other side of a firewall built into the C3 Agentic AI Platform. As such, all queriers—along with all processing and data analysis—happen within your enterprise systems without connection to the internet. The result is zero risk of IP or data exfiltration by the LLM.

5. Free of Hallucination

As ChatGPT and other LLMs have soared in popularity, so has attention to the fact that its seemingly remarkable answers are sometimes flat-out wrong. The LLMs that interpret a user’s queries don’t have a good way to check the accuracy of source data, nor are they designed to tell you when they don’t know an answer. Similar to the iPhone keyboard’s predictive-text tool, LLMs form coherent statements by stitching together data — words, characters, and numbers — based on the probability of each piece of data succeeding the previously generated piece of data.

When LLMs do not know an answer, they frequently make one up. The results are pure fiction – technically known as hallucination. Any Generative AI solution for the enterprise must prevent hallucination.

Instead of relying on an LLM to generate an answer, the LLM should effectively hand off the query to an underlying orchestration agent that retrieves the answers from deep learning models already applied to an enterprise’s data. C3 AI solves this by not relying on the LLM to provide the answer. With C3 Generative AI, the LLMs have no direct access to the data. The LLM is enabling the enterprise AI software already applied to an organization — and thus provides reliable responses. Even better, an ideal generative AI system for the enterprise should tell a user when it doesn’t know an answer instead of generating an answer strictly because that’s what it’s trained to do. C3 Generative AI only provides an answer when it’s certain the answer is correct. If it can’t find an answer, it tells you so. Deterministic answers. No hallucination.

6. LLM Agnostic

Importantly, C3 Generative AI is LLM agnostic. That means C3 AI can use whichever LLM is best suited for a particular customer or industry, and we can take advantage of the rapid innovation in LLM development. If a new LLM emerges that has distinct advantages, we can incorporate it into our architecture. We also fine-tune LLMs so they can support highly secure environments required by some customers.

C3 AI supports Azure GPT-3.5, Google PaLM 2, AWS Bedrock Claude 2, AWS Bedrock Titan, C3 AI Fine Tuned Falcon-40B, C3 AI Fine Tuned Llama 2, and C3 AI Fine Tuned FLAN T5.

How C3 Generative AI Is Transforming Businesses

One major paper manufacturer is now using C3 Generative AI to significantly boost the speed at which it can address maintenance issues, ensuring its machines run as efficiently and productively as possible. The benefits far surpass what’s possible in a pre-generative-AI world.

Just one paper machine can stand eight stories high. It can contain several thousand sensors producing billions of records of data daily.

The company faces increasing challenges as longtime employees retire, taking years of valuable knowledge and hands-on experience with them. C3 Generative AI addresses that issue by making it possible for junior workers to easily get answers to routine or complex problems.

Now, an operator at the plant can ask simple questions, and get fast, helpful answers.

The system then provides a step-by-step method to fix the issue.

This is a monumental breakthrough compared with how things currently work at most manufacturing plants. While the process of maintaining and monitoring manufacturing equipment today is typically managed by an experienced machine technician, the process can be daunting. The technician needs to know which PDF manuals to search through — operating manuals from the original equipment manufacturer (OEM), numerous standard operating procedures (SOPs) written by the operating company itself, plus multiple spreadsheets with logs of sensor data. Ultimately, if an urgent problem arises, the team likely has a go-to person (or people) that know how to address the issue the fastest.

C3 Generative AI simplifies such laborious and time-consuming processes. The machine operator types the query into the application, and the generative AI system does the rest, retrieving and delivering the most useful information possible in a matter of minutes versus hours or even days.

These sorts of process efficiencies are poised to transform enterprises everywhere, across multiple industries – financial, retail, hospitality, healthcare, defense, oil & gas, and more.

Consider the workflow for a sales manager at a financial service firm. Sales managers spend 45% to 60%¹ of their time searching through product documents instead of focusing on selling. C3 Generative AI can eliminate the bulk of that, allowing sales managers to ask and receive reliable answers about the firm’s product offerings within minutes. An analyst at that same firm who needs information about a bond offering can use generative AI to rapidly retrieve important information about risk, maturity dates, and more; a loan manager can quickly find out which customers are at risk of default in the next quarter.

It’s hard to imagine an industry or government agency for whom generative AI won’t lead to important, transformative changes. If a job involves getting answers from documents and numerous other data sources, from sensors, or data residing in data lakes and various applications – generative AI is a game-changer for the enterprise. The possibilities are limitless, and the implications for accelerating and improving digital transformation efforts, impossible to overstate. This phase of the digital revolution is happening so fast that those who wait will fall woefully behind and, to put it bluntly, risk extinction.

See how C3 Generative AI applies to your enterprise data. Explore the product, or schedule a demo to see it in action.

¹ Estimates based on customer projects.