We're replacing our current helpdesk HaloITSM this quarter and I'm surprised by how many options there are now.
Looking at Freshservice, Jira Service Management, Zendesk and monday service.
If you were starting from scratch in 2026, what would you pick and why?
Our company has been using separate tools for customer support and internal IT for a while now, and it's starting to create more work than it saves. requests end up getting forwarded between teams, reporting is split across different systems, and it's hard to get a complete picture of what's going on.
We're considering moving to a single platform where both customer facing support and internal service requests can be managed together, with shared workflows, automation, and reporting.
Has anyone gone this route did it actually simplify things, or did you end up compromising on features that were better in dedicated tools?
Ok so for context im the IT service manager at a mid size company and we just rolled out this shiny new enterprise service management setup to centralize IT and HR ticketing workflows in one place. Think unified portal, one set of forms, shared automations, all that good stuff.
We built a single service portal so employees see one entry point and it routes to either IT or HR boards. I was trying to be clever and created a global category called Access then tied a bunch of automation rules to it. HR uses the same portal for things like promotion letters and comp changes. Guess who reused the same Access category for both IT and HR requests and then wired it to one master notification board.
Launch day, people submit tickets for VPN, laptop, and also salary adjustments and performance issues. Because of my config, the “All Requests” view was shared to IT, HR, and a few business team leads as a fancy dashboard demo. Every ticket description and attachment showed up there. So for about 4 hours, random managers and some senior engineers could see tickets like “Please adjust Jane’s salary to X” and “We need to put Bob on a PIP” sitting right next to “reset my password”.
HR pinged me like WHY CAN ENGINEERING SEE COMP REQUESTS. I feel so embarrassed. I killed the view and ripped out the automations but damage is kinda done, people already saw stuff they should not have. Trying to spin it as a misconfigured filter in a new ESM rollout but omg this is rough... appreciate any thoughts
Looking at this from the transport side rather than the security side.
AI traffic has a different profile than the web traffic our current design was built around: long-lived connections, streamed responses, bursty token throughput, and a small number of destination endpoints concentrated in a handful of US and EU regions. Add inline inspection on top and you are doing decrypt, inspect, and re-encrypt on sessions that stay open for minutes.
What that surfaces: PoP density matters less than PoP placement relative to model endpoints; inspection adds compute at the edge, which not every PoP footprint is provisioned for; our sites in South America and Southeast Asia are already the outliers on RTT, so adding an inspection hop there is the thing I am worried about; and peering to the major cloud and AI providers varies a lot by vendor and nobody publishes it clearly.
Questions for anyone who has measured rather than trusted the marketing map: what is your real added latency for an inspected AI session from a secondary-market site; does the vendor do inspection at every PoP or only at a subset of "full-stack" locations, since several are quietly the latter; and has anyone hit throughput ceilings on inspection during peak?
Our support team has been getting slammed lately, so management has been pushing AI pretty hard to help with the workload. I get why nobody wants to spend all day answering password resets, access requests, and the same basic questions over and over.
What worries me is giving AI too much control too quickly. We've already had cases where a ticket was routed to the wrong team because the bot misunderstood the request, and fixing that ended up taking longer than if a person had handled it from the start.
I'm all for using AI for repetitive stuff, but I still want someone to be able to see what it did, why it made that decision, and step in when something doesn't look right. I don't think our team wants full automation we just want less manual work without creating even more cleanup later.
How are you all using AI in your service desk? Has anyone found a setup that actually saves time without making everyone nervous about what the bot is doing?