You need to be very clear about the persona who you're building for, what their pain point is, and why they're willing to spend money to solve it. So far it seems like you took an emerging technology (agentic workflows), applied it to a novel area (DevOps), built a UX around it, and tried to immediately start selling. This is the product trap of a solution in search of a problem.
Are you trying to sell to large companies? The problem that large companies have is cultural/organizational, not tooling. For any change, you need to get about a dozen people to review, understand, wait for people to come back from vacation, ping people because it fell off their desk, sign off, get them to prioritize, answer questions again from the engineer the task was assigned to, wait for another round of reviews and approvals, and maybe finally somebody will get the fix applied in production. DevOps is (or at least, it originally used to be) focused on finding and alleviating the bottlenecks; the actual process of finding data or applying changes is not the bottleneck in large companies and so therefore it is not a solution to the pain point that different folk in large companies have. If your value proposition is that large company executives could replace Infrastructure employee salaries with a cheaper agentic workflow, you need to re-read my prior point - if large companies have all this process and approvals for human beings making changes, why would they ever let an agentic workflow YOLO the changes without approval? And yes, I know, your agent proposes Terraform PRs for making changes to keep a human in the loop - but now you slayed one of the Hydra's heads and three more have popped up in its place: the customer needs the Terraform PR to be reviewed by a human committee, some of whose members are on vacation, some of whose members missed the PR request because they had other priorities and it fell off their desk, etc. etc. Doesn't really sound like you solved anything. The fundamental difference between what you built and something like Claude Code is that Claude Code doesn't need a human committee to review on every iteration it executes on an engineer's laptop, only the review of the One Benevolent Laptop User who is incentivized to get good output from Claude Code and provide human review as quickly as (literally) humanly possible.
Are you trying to sell to small companies that don't have DevOps Engineers? What's the competitive space here? The options usually look something like, (a) pay a premium for a PaaS, (b) spend on the salary for your first DevOps Engineer in the hopes that they will save more on low-level infra bills compared to their salary, so you're posing now (c) some kind of DevOps agentic workflow that is cheaper than a DevOps Engineer salary but will provide similar infra cost savings? So your agentic workflow will actually lift and shift to better/cheaper infra primitives and own day-to-day maintenance, responding to infra issues which your customers - who aren't DevOps Engineers, and don't know anything about infra, and are trying to outsource these concerns to you - which your customers don't know how to handle? I would argue that if you really did achieve that, then you should be building an agentic-workflow-maintained PaaS that, by virtue of using agents instead of humans, can undercut traditional PaaS on cost while offering a maybe better UX somehow. If you're asking your customers to review infra changes that they don't understand, then they need to hire a DevOps Engineer for the expertise to review it, and then you have a much less interesting value proposition.