What needs improvement with CircleCI?

Debugging is an important point for improvement because when pipelines fail, debugging can sometimes be difficult, especially for the complex workflows which we have already faced in our organization in different projects. Logs are available, but root cause analysis can take time, so that can be improved. Also, configuration management challenges exist. Although YAML configuration is flexible, it can become complex and hard to maintain as pipelines grow in size and include multiple workflow conditions. The limited visibility for large pipelines is also a concern. For very large enterprise pipelines, the UI can feel less intuitive in terms of tracking dependencies and understanding the full workflow at a glance. Cost optimization is one important point for improvement. In cloud-based usage, costs can increase due to long-running jobs and inefficient pipelines. Proper optimization through caching and parallelism is required, otherwise it can become expensive, which we faced in one of our telecom projects last year. One thing CircleCI can improve is related to cloud runners, where any network or platform issues can impact pipeline execution. This can be a concern for critical telecom deployments.

Regarding flexibility, Azure pipelines has the ability to pass variables with greater flexibility than CircleCI offers. CircleCI has this feature, but the flexibility in Azure pipelines is something CircleCI does not match. We pass runtime values such as secrets or API values, and this capability could be improved.

One limitation I see in CircleCI is the troubleshooting of complex pipeline failures, which sometimes takes time, especially when multiple jobs or containers are involved. More intelligent root cause insights would be helpful. The configuration experience often depends on YAML setup, which feels more technical for new users not from the DevOps team. A more guided visual pipeline builder would ease onboarding. I also feel that deeper flaky test analytics would add value because quickly identifying unstable tests versus actual product defects is important for QA teams. Integrations are generally strong, but sometimes teams need more plug-and-play connectors for niche tools or simpler setup steps for third-party testing platforms. Making those integrations more seamless would save onboarding time. The technical content in documentation is useful, but in some advanced scenarios, it can take time to find exact solutions. Including troubleshooting guides or real-world examples would be helpful. Regarding support, faster resolution for urgent pipeline blockers is always valuable, especially when builds impact release timelines.

I would suggest some advancement around AI, integrating it with CircleCI, as I see some initial changes happening, but it could help with Agentic AI workflows, allowing us to build our Agentic workflows within and with CircleCI, which would be really helpful as most applications are focusing on Agentic AI. More features and usability around Agentic AI would be really helpful, but otherwise, things are good. I do not have anything else to mention as an improvement needed for CircleCI.

As each time code is deployed onto the main branch, the build automatically triggers, saving us time. We have reduced our manual efforts significantly after the initial setup. We used to spend around four to five hours per week managing deployments and related processes, which has been reduced substantially. CircleCI is fairly simple to use and set up. Sometimes the documentation is too in-depth. There are many features, but I think sharing the top ones would help end users set it up more quickly and efficiently. Simpler documentation and implementation highlighting the top features instead of presenting a huge bulk of information would make the experience perfect.

CircleCI can be improved with more integrations with different cloud platforms and by providing environment security features. It could provide integration for secrets with Vault or other security tools.

CircleCI can be improved by making it less costly, as it is very expensive. The config complexity, like the YAML config, can become messy in complex projects. Making it simpler, much like having a Docker Compose YAML or Kubernetes YAML, is necessary from that perspective. Rather than keeping it a SaaS project, they can think of it through a Jenkins approach, where we can also self-host it into our environment, but it is acceptable. It is very expensive, and many organizations cannot afford it. The config complexity, like the YAML configs, can become messy in complex projects. A better DevOps person can only handle it, not a normal person. For that reason, I chose a rating of seven. It is quite expensive, to be honest. As mentioned, many organizations cannot afford it because of the parallel execution prices as well as the config complexity.

I think CircleCI can improve in terms of pricing and needs to bring more unique selling points to stay competitive as there are many other cloud solutions such as GitHub Actions and Azure DevOps that offer similar execution and deployment services.I believe CircleCI should leverage the growing trend of AI by offering an out-of-the-box AI-driven dynamic YAML file creation feature using natural language processing. Instead of having users define pipelines manually, users could simply use natural language to define their needs and let the AI generate the YAML configuration automatically. Implementing a multi-agent architecture could also enhance usability. I rated CircleCI six out of ten because I think they need more transparency in pricing, as there are instances of unclear network data transfer and storage costs related to caching and workspaces. Additionally, there are sporadic platform security incidents that need addressing, and pipelines require good protection. Secret rotation should also be prioritized. There are occasionally long-running workflows that can be difficult to debug, and improvements could be made in job startup times and cache management.

CircleCI's Orbs are not as easy to use as they could be. If CircleCI implemented a function approach similar to GitHub Actions, it would be significantly better. CircleCI could be improved if it could function without Orbs. Instead of Orbs, using a function approach like GitHub Actions would be beneficial.

We've had occasional connectivity issues with cloud resources and build failure due to its own internal system setup and environment. That costs us credits. Support engineers do not thoroughly read and understand issues when emailing them. They reply to me with a totally different context about the problem. Just yesterday, my web applications failed to 'yarn build' after I downgraded the resource class from extr-large to large. I contacted support and got no reply. Billing is a mess. There is duplicate information in the bill downloaded from their website and when I asked for a consolidated bill, they just answered that they didn't have it. There is no transparency on how many billed minutes and credits I spent each day both on their website and the monthly bill which makes it difficult to understand when the monthly bill contains many refill item charges. It is difficult or impossible to track usage and burndown of my subscription and to gain total outstanding refill amounts on a daily basis. Their website doesn't provide sufficient information on credits allocated/calculated for my extra-large resource class.

The solution’s pricing could be better. In future releases, it may provide integration with Grafana or any tool for visualization.

There needs to be some improvement in the user interface of CircleCI.