C3 AI reviews

In some sense, C3 is aptly named because it shares several similarities with C-3 (the predecessor to C-4): extreme toxicity, potential to go up in flames, and eventually obsoleteness. Here are some targeted problems plaguing C3: *Fundamentally flawed business model* A software company typically starts with some kind of mission and a somewhat defined problem to solve. C3 started with a list of sales contacts and a mountain of cash. This is an enviable position for any startup, but C3's mission never got more specific than solve any enterprise problem for any customer we can get. [Correction, now the company mission is "C3 AI everywhere, gen AI everywhere"]. Terrific! After all, how different can an oil/gas enterprise be from a food processing corporation? The result is limited personnel scrambling to build dozens of completely unrelated software projects, many of which become wasted effort from natural sales attrition. This is obviously not scalable, so the next big idea is to cluster some loosely related customer projects and build base applications. Sounds like a good idea in theory and works to some extent in a couple cases, but because of the huge diversity in enterprises (even across factories in the same business), the concept of base applications are actually a burden since field teams need to figure out not only what to build for a customer, but also what needs to be changed from a base application (on average, many months of several full-time employees' time). The official sales lexicon for a customer project is we "configure an application", but the translation in reality is even worse than “build from scratch" for many projects. And that is how we end up with an unfocused array of "130+ turnkey applications". If C3 chose to focus on a few high-value use cases, they might have had a better chance of building a software consulting firm, but there are more reasons why C3's future is bleak. *A sunk-cost platform* C3 AI's platform was an interesting idea in the early 2000's. While OOP taken to the extreme with databases and task distribution is a theoretically interesting toy project, it is now 2025. Not only has C3 itself rebranded several times to pursue hot buzzwords (C3 Energy -> C3 IoT -> Gen AI -> Agentic AI), but technology itself has also changed drastically. Despite all of this change, C3 stubbornly clings to its outdated platform because it would be a shame for leadership to drop such a massive investment. A software-consulting business model requires a platform that enables flexible and fast development due to constantly shifting project requirements. Unfortunately, the dense interconnectedness of different (often unwanted) components stemming from the type system results in chaotic errors only highly skilled people can debug, hours spent on tasks that can be done in minutes in other companies, and the most inflexible and slow UI framework I have ever seen. Not to mention poor software performance due to overhead from needless wrapper layers and generally poor system design that causes devastation for any customer projects with even slightly demanding workloads. Meanwhile the platform team keeps being forced to work on useless features from the whimsy of our CEO. The data science teams are even so frustrated with the platform, that they are each independently spending effort to build basic tools to bypass platform nuisances. However, solutions and apps engineers don’t have these luxuries even though development would be easily 10x faster without the platform. *Career endangerment* For those in technical roles (software engineers/data scientists), staying too long at C3 can be detrimental to your career. Because of C3’s unique tech stack, you will only know how to use C3 “tools” as long as you are there. Especially for those working on customer projects, there is virtually no opportunity to work with transferrable skills other than basic JS and Python. Most data science work doesn't even involve more than an Intro to Machine Learning class. Additionally due to the high-stress nature of customer work in tight timelines to meet revenue targets, engineers and data scientists often take shortcuts so that the code you see is exclusively technical debt and would be grounds for performance-based firing in another company. This effect is manageable for those early in their career (new grads and mid-level’s) and for platform engineers, but many who have been at C3 for several years eventually realize the magnitude of their opportunity cost in technical growth from staying at C3 so long and spend significant time to catch up on new skills before interviewing. *Fear culture* Executives and managers need to succumb to all of our CEO’s demands and treat his word as the gospel; otherwise, they are gone within the day. The CEO has made it clear that since we are “professionally paid”, it doesn’t matter if we “break our backs” to deliver on his requests. This fear chain trickles all the way down: e.g. applications engineers especially frequently work long hours on weekends to meet artificial, frivolous deadlines and can even be considered for a story point of work during a serious leg surgery if the surgery only takes a few hours. A side effect of C3’s fear culture is a big telephone game in communications across C3’s company hierarchy. Depending on your level in the company (and even the floor you sit), you may have a very different perception of the company from someone else. Consequently, the CEO is completely oblivious to the reality of C3’s products and customer projects. He certainly doesn’t seem to grasp how little C3’s Generative AI application really adds to the LLM’s everyone uses. There is a huge exodus of talent from C3, especially the experienced heavy-hitters. The remaining people are either new grads, contractors, or people with visa restrictions. Pressure from Wall Street to increase revenue in tandem with a net loss in experience over time creates an environment reeking of desperation where people are in survival mode every day. If you have the patience to 1. deal with swarms of ex-McKinsey leadership who aggressively don’t know the first thing about software/ML, 2. work with a burdensome and irrelevant tech stack, and 3. explain to customers why they should pay for unsophisticated, low-value software, then by all means join the circus. Otherwise, especially if you have a technical background and want to stay relevant, don’t do this to yourself.