Companies at this scale are really a group people and a culture and not a nebulous entity. This team has been fantastic in communication, issue resolution, and proactive attempts to identify problems. They have lived with the problems of not having this tool so they understand what needs to exist to have a good end-user experience.

Can see FW deployments at a glance Can create sub-groups to deploy specific FWs to for internal testing (And can change on the fly) Better stacktrace visualization than my IDE Real-time visibility into new problems and regressions Has helped me track down stack overflow issues, heap fragmentation issues, issues with the manufacturer's BLE stack and all the kinds of critical hard-to-track problems that would have taken 10x longer without it. There is nothing else like it. It eases debugging of any problem more than any other tool I have ever used.

(Note: these are all minor and I have already notified the Memfault team about them.) I can't easily get access to the raw per-device metric data via a documented API. I can't easily see the FW version that the metric data was produced on via the web interface. There is a bit of confusing about which visualization tool to use for the metric data and it seems to have some bugs. It is hard to manage the metrics generally. There is a limit on them it is hard to deprecate ones you create that you realize you don't need. Etc. Some confusing things about being able to deploy debug/release builds and managing variants of build types. (Perhaps more education on our part could have fixed this?)

When deploying huge numbers of connected products, it is an inevitability that some of them will experience faults and crashes. The product we integrated Memfault into is battery operated so every fault incurs a hit to the battery life, which could ultimately lead to the product violating its warranty, creating big problems for us. Without the functionality offered by Memfault built into the product those issues can be all but impossible to ever resolve. I am relieved that we are using the service as it gives me confidence that those issues, when we encounter them in the future, can be solved quickly and effectively.

With my experience as an IoT firmware engineer I did not struggle to integrate Memfault into our existing product at all. Memfault support was fully available as well to help answer the few questions I did have. Using their webapp to inspect faults and their respective traces is very easy. Their webapp is easy to use and intuitive, and guides you quickly to getting immediate value from the data it reports.

Integrating the product does require at least a firmware engineer, if not someone on the Cloud side to handle the messages from devices.. Memfault does a lot to make it easy, but that could be a high barrier to entry if a company doesn't have those resources on staff already.

Amazing when doing trial deployment to be able to diagnose issues fast remotely. Firmware issues are now so fast to find!

Amazing for managing firmware versions and examining what is happening with individual units.

There is limited API access to the data. And custom queries on the data can be hard to do.Fleet graphs are somewhat limited.

Memfault gives us visibility into how our devices/firmware is performing in the field. Their support is top-notch.

Memfault was very easy to set up with our nRF52 MCU. Their SDK is excellent and plugs right in. It allows us to track device metrics and various errors. Without Memfault we would either have no visibility into what's happening on devices or have to build an in-house solution. One thing I really like about Memfault is the ability to get coredump details (including file and line numbers).Memfaults OTA functionality is great as well. They handle all the business logic of whether the device needs to be updated and serves new firmware when necessary.

The fine grained access control is lacking. Every user has access to nearly all of memfault. This can be problematic for production deployments when you want only a select group of people access to something (eg: roll out OTA updates).

Memfault is easy to integrate and scales without any additional effort. It gives our startup hugely valuable capabilities while still being really accessible.

Memfault's data visualization still seems to be in it's early stages, but have noticed this improving with time. Not a major factor for us though. Overall, no complaints.

Memfault can help with not only your deployed fleet, but also in-house debugging. It's extremely useful for easily displaying all information involved with crashes in a very user-friendly way, instead of having to manually parse out data with scripts on the command line.

The pricing is a bit opaque. They have a reasonable explanation and are super willing to work with you on getting a flexible plan that works for you though.

Memfault support is excellent and I liked it the way they made themselves available for us during whole integration.

Ability to record a software issue (e.g. especially program crash and watchdog reboot) on device and make necessary information available on cloud.

Inability to provide full information to developers for C++ project. Cloud interface could be more user friendly

Memfault helps to create custom metrics to track the HW behavior and also provides a timeline to look at HW performance, this is the best to understand, compare and improve the HW performance.

Data visualization can be better but memfault takes the reviews periodically and improve their product as well.