Enterprise agents: what’s working, what’s still painful, and what changes in 1–2 years?
retained text
We work on software for operating agents, and we’re gathering feedback from our teams, colleagues, and developer communities on what enterprise deployments look like in practice. There seem to be plenty of agent developers here, so let’s compare notes. Getting a workflow into daily use brings the details into focus: output quality, access to company data, integrations, handling exceptions, and how much time people spend reviewing or fixing results. Those details affect whether the business actually saves time or just moves the work around. I’d like to hear your experience: * **What’s working?** Share one workflow, the business problem it addresses, and what has actually improved. * **What’s still painful?** What takes the most effort to get right, and has anything meaningfully helped? * **What changes over the next years?** What do you expect companies to routinely trust agents with, and which expectations seem ahead of reality? Pick whichever question you have experience with—no need to answer everything. Successful deployments, stalled pilots, and cases where simpler automation worked better are all useful. A specific example would help everyone compare experiences. Rough industry and company size are enough; no company names needed.
document fetched 2026-09-07 00:18 · canonical https://reddit.com/r/AI_Agents/comments/1w92eb2/enterprise_agents_whats_working_whats_still