What is missing, said out loud
This is a list of gaps rather than a list of ambitions. Every item is something the product genuinely does not do today, written so an evaluation can be accurate rather than optimistic. There are no dates on it: a dated roadmap is a set of promises, and promises about unbuilt software are the cheapest thing a vendor can hand you.
Blocking a procurement conversation
These are the ones that stop a deal outright rather than costing a feature comparison. If any of them is a hard requirement, the honest answer today is no.
Compliance certifications
No SOC 2, no ISO 27001, no HIPAA and no business associate agreement. These cost real money and real elapsed time, and claiming to be "working towards" one is how vendors imply a certificate they do not hold.
SCIM provisioning
Single sign-on is live over both OpenID Connect and SAML, but users are not provisioned or de-provisioned automatically from your directory.
A choice of region
Data is processed in Germany and that is the only option. There is no European multi-region choice and no other continent.
Formal invoices
Payment works and a receipt is issued. A formal, downloadable invoice of the kind a finance department files is not yet, and it is a tax-registration question rather than an engineering one.
Features a comparison will find
- Continuous profiling and front-end profiling. Neither is built.
- Bundle size analysis.
- Ownership and code-owner rules, for routing an issue to a team without a human step.
- Advanced inbound filtering, beyond the substring and exact-match rules that exist.
- A first-class view of agent and tool-call spans. Such spans would land today and be stored; what is missing is the read path and the page.
- Named one-click Google and GitHub sign-in buttons, rather than reaching both through OpenID Connect.
- An on-premise event proxy, and a self-hosted build of any kind.
Deliberately not planned
Some absences are decisions rather than backlog, and treating them as backlog would be dishonest about what this product is trying to be.
Failure prediction
Predicting an outage that has not happened requires guessing. Assessing the regression risk in a change that has been written is the defensible version of the same idea, and that is what code review does.
A cost calculator
Calculators exist because usage is billed. Pricing here is flat, so a calculator would either print the same figure whatever you moved, or imply a usage bill that does not exist. A plan chooser answering "which one do I need" is the honest version, and is worth building.
Billing for overage
Usage over a plan is announced, not charged and not dropped. Reversing that is a business decision with contract changes attached, not a feature.
Opening pull requests for you
A fix comes back to your pipeline rather than being pushed. Holding a write credential for a customer’s repository is a blast radius this product does not need to take on.
Ask for one
This list is ordered by our judgement of what matters, which is a poor substitute for yours. If something on it is blocking you, saying so moves it — there are few enough customers today that a single clear requirement genuinely changes the order.
Tell us which one is blocking you
Write to [email protected] with the requirement and the deadline. A specific blocker from a real evaluation outranks anything on this page.