I have been using C3 AI for five years. My main use case for C3 AI is predictive maintenance, which belongs to reliability asset in development.

A specific example of how I use C3 AI for predictive maintenance is that we have different assets coming through as predictive maintenance related to pumps, compressors, and other equipment. C3 AI helps with those compressors and other equipment by spotting issues before they happen and reducing downtime, which is exactly what we used to have a use case for.

When we use C3 AI for predictive maintenance, we raise a critical alert when we identify any exceptions, helping us to identify any preventive issues and reduce shutdowns.

I have noticed improvements in efficiency, reduced downtime, and time saved for my team, as it is quite integrated through the source ML pipeline to report to general customer UI, functioning as one platform that takes care of everything, thus improving efficiency, optimization, and increasing project deliverables.

We compared what we had earlier with what we got from C3 AI, and it is an actual improvement that helps us identify the KPIs from an AI perspective. C3 AI has positively impacted my organization by providing optimization on the project.

The best features C3 AI offers are its reliability, robustness, and different connectivity with different source systems, as well as a large amount of AI implementation and available libraries and workflows. All of these features are interconnected to the predictive maintenance model to get the result.

C3 AI seems quite understandable and has unique cases with less code required, which is beneficial.

I noticed that C3 AI's support is primarily handled by the operational team, and if you are new to C3 AI, you will not find available information in searches, which is limited. Support should be available in global training and documentation for beginners.

Regarding documentation, it should be more readily available. The reason I chose eight out of ten is that as a user of C3 AI, I found the operational aspect not easy to understand; it is a bit tricky, and you should have more core knowledge to navigate C3 AI.

C3 AI is stable.

Regarding C3 AI's scalability, I think it is good; the only issue is with the clustering aspect, such as sharing a cluster, but the scalability is very high.

Customer support is good, but as I mentioned earlier, there is significant dependency on the operational teams from C3 AI, so as a third-party consumer, we do not have much authorization from our side to take care of infrastructure and back-end things.

Before choosing C3 AI, we did not browse other options.

My advice to others looking into using C3 AI is that it is good, and if you want to have a stable and scalable product with broad use cases available, C3 AI is a reliable application suitable for your use case.

I want to provide a suggestion as feedback: C3 AI should take care of the infrastructure, cluster optimization, and scalability, and also think about customer support as it relates to the operational team. If those things improve, then C3 AI is a good product. I rated this review eight out of ten.

My main use case for C3 AI was as an implementation partner working for a technology consultancy firm, where we attempted to introduce C3 AI to our clients, primarily in the financial services sector.

We wanted to introduce C3 AI for three reasons. First, it had been funded by Tom Siebel and his son, and given the reputation and legacy of the Siebel family, we believed they have strong connections in the corporate software world. Siebel and his legacy CRM solutions, along with his network, are important assets. Second, they wanted to bring enterprise-level engineering and software building to the world of AI, which was very attractive to us because in many cases, AI consists mostly of proof of concept and small experiments that never reach large-scale applications. If C3 AI could make that work and reach the enterprise level, that would be really good. Third, the main interface for developing anything with C3 AI is based on JavaScript, meaning that many traditional corporate IT professionals could also work on AI implementations without needing deep training in specific machine learning languages such as Python or R. This was the primary reason we wanted to work with C3 AI.

One of the best features C3 AI offers is that you do not need to learn Python or any specific machine learning programming language; you can use C3 AI with JavaScript. The main coding development interface is JavaScript, meaning there are many more people worldwide who can work with it.

The impact on my teams and clients is that it creates interest among many individuals; however, as you go deeper, people realize that doing AI is not just about programming and building the software. You need to understand machine learning models to some extent, so ultimately, it does not yield the positive outcome expected. You cannot expect a JavaScript developer to overnight become a data scientist without knowledge of machine learning models. At first glance, it appears very attractive, but if you dig deeper, it does not work as effectively.

If I could change or add anything to improve C3 AI, I would suggest making it not just an AI platform, but a data and AI combined platform, as bringing in data management elements, governance, and cloud features is critical. The real AI and modeling represent only 20 percent of what is needed; there is over 80 percent in the rest of the world that should be addressed. Given Tom Siebel's deep understanding of corporate work and his network, integrating C3 AI with other ERP solutions such as Oracle, SAP, or Microsoft is essential. If C3 AI can be an enterprise platform, embedding more data elements and ensuring seamless integration with other ERP solutions would significantly enhance its capabilities and attractiveness. If someone told me that they have SAP, and C3 AI can seamlessly integrate the data and perform action calls between SAP, allowing me to operate on C3 AI as I would on SAP, that would be a killer feature for me.

I would suggest avoiding traditional licensing sales as it is too expensive; the starting price is too high. Considering a subscription model or a pay-per-use model is the best business structure for AI or data solution vendors. The pay-per-use model benefits usage, as the more you use, the more successful you become with the platform.

I have been using C3 AI from 2019 to 2021, which totals two years.

Based on my experience, I do not see C3 AI being used to bring AI projects to enterprise scale at my client. I can provide some real examples. We were discussing a big bank in the city of London, which is progressing to the cloud platform and GCP from Google, and Google has offerings very close to themselves, such as BigQuery and Kibana. They are hesitant to switch to a totally different platform for AI projects. In another instance, I spoke with a CTO of a leading consumer goods company, who mentioned that C3 AI solutions are too expensive and that he cannot see clear use cases before deciding to invest in C3 AI. Even for this specific use case, it still does not provide him with very clear and tangible business case value. How could he skip the experimental stage and directly spend large amounts of money on a platform? Additionally, when comparing C3 AI to another emerging vendor, Databricks, I find their business model to be much smoother; they provide a platform solution for free and charge by usage. They call it data break points, which is probably a better and more appealing business model for CTOs deciding on an AI platform. When discussing AI, you also want the data, so it is critical to have a data-AI combined platform. So far, we find Databricks quite good, but C3 AI as just an AI platform does not make that much sense.

I have not seen a return on investment with C3 AI, and I cannot share any.

From my experience, I see the pricing, setup costs, and licensing of C3 AI as quite expensive; even a CTO from a globally leading consumer goods company feels it is too expensive and does not want to give it a try.

Before choosing C3 AI, we evaluated other options such as Databricks and Snowflake.

You do not want to use C3 AI for proof of concept; you want to see C3 AI help you bring something to enterprise level at large scale. In that aspect, I never see it happen based on my experience.

We even failed to open any doors for C3 AI; there are no smaller wins, learning opportunities, or process improvements, even if the main goals were not met.

My advice to others looking into using C3 AI is to ensure that if you really do not want to hire data scientists, and your organization predominantly consists of traditional software engineers who mostly know JavaScript, you should go for it and give it a try.

I would give C3 AI a rating of four out of ten because I have not seen any significant impact from C3 AI so far, and there is no real evidence or strong, large-scale, successful use cases in front of me.