Send every ticket to the right team in under a second.
NeoOps learns from tickets you already resolved, then tells you who should own a new one, which RCA agent to run, or any other field you define. No model to train. No GPU bill. A fraction of a cent per incident.
- INC4274218Customer reports order arrived damaged, wants refund→ Customer Care · Returns0.9343 ms
- INC4266299Volume /data reached 95 percent on vm-04→ Storage Operations0.97Disk Capacity Agent96 ms
- INC4258380AP client utilization (72) exceeded critical threshold→ Wireless Global Support L30.84Wireless Capacity Agent59 ms
- INC4250461Active SSL VPN sessions below threshold, currently 0→ Network Security L30.88User Connectivity Agent112 ms
- INC4242542Host resource: CPU exceeded 70% threshold on srv-27→ Server Operations L20.91High CPU/Memory RCA Agent75 ms
- INC4234623Interface Gi1/0/9 changed state to operationally down→ Network Enterprise L30.96Interface Flapping Agent38 ms
Every ticket to the right team, the first time.
Your resolved tickets show where each new one belongs.
A wrong first assignment costs more than the ticket itself.
Triage is still mostly a person reading a ticket and guessing a queue. When the guess is wrong, the clock keeps running while the ticket waits in the wrong place.
of tickets in manual triage land with the wrong team first.
[1]added to resolution time for every reassignment.
[2]of the requester's work time lost per bounce. Four bounces cost over nine hours.
[3]A Level 1 fix averages $22. Dispatch it to desktop support and it costs $62 more.
[4]Well-run service desks aim to keep reassignment under 10% [5]. NeoOps is built to get you there without new rules, new headcount or a model project. It is not limited to IT either: any queue of cases with a history of who handled them works the same way, from customer support to billing disputes.
Two steps to your first routed ticket.
No rules to write and no model to train. Your resolved tickets are the examples, so a new team or category takes effect the moment it shows up in your history.
Import your resolved tickets
Upload a CSV from your ticketing system. Fields are mapped automatically. Review or customize the mapping during import.
Send new incidents to the API
Call NeoOps from your workflow or agent. Get the predicted team or field value with a confidence score, then assign it automatically or send it for review.
One API for the decisions your L1 team makes all day.
Each question is a field you configure. Assignment group and RCA agent come ready to use. Any other field in your tickets works the same way, as long as your history has it filled in.
assignment_groupWhich team should own this?
rca_agentWhich agent should run the root cause analysis?
any_fieldAny other field you have history for
A fraction of a minute and a fraction of a dollar.
General models can route tickets. They are slow and expensive at it, because they read long prompts and write long answers for a decision that needs one label. Here is a 1,000-incident queue run four ways.
Simulation of a 1,000-incident queue using throughput, accuracy and per-incident cost from our internal test runs against hosted LLMs. "Needs attention" means the predicted group was wrong. A full public benchmark with method and data is coming.
What NeoOps is
- A model built for one job: picking the right value for a ticket field.
- Grounded in your own resolved tickets, read at request time.
- Fast enough to sit inline in ticket creation.
- Honest about doubt. Every answer carries a confidence score.
What it is not
- A chatbot. It does not write summaries or replies.
- A model you train, host or tune.
- A wrapper around someone else's LLM.
- A replacement for your ITSM. It fills fields in the one you have.
Your tickets stay yours.
Zero retention on routing requests. No training on your data. The history you upload stays in your account, and you decide how long.
$1 to start. Then pay per call.
Sign up with Google or email and you get a dollar of credit, which covers on the order of a thousand routed incidents. When it runs out, reload from $5 with a card. No seats, no contract, no minimum.
Running a large queue, or need it inside your own cloud? Talk to us about volume and private deployment.
- Typical cost per incident~$0.0008
- Billingper token, shown in your account
- Reloadfrom $5, any time
- Credit expirydoes not reset monthly
- Web playground and APIsame credit
Before you upload anything.
Do I need to train a model?
No. You upload resolved tickets and NeoOps uses them as examples at request time. Change your teams and the new history takes effect on the next request.
How much history do I need?
Start with a few hundred correctly labeled tickets. A smaller clean set beats a large set with missing or lopsided labels. The readiness score shows which fields have enough to work with.
Does it only work with ServiceNow?
No. Templates cover ServiceNow, Jira Service Management and Zendesk. A custom column mapping covers BMC, Salesforce cases or a homegrown tool. Each account works with one schema.
What happens when NeoOps is not sure?
Every answer has a confidence between 0 and 1. Pick a threshold. Above it, assign automatically. Below it, send the ticket to your triage queue as you do today.
Do you train models on our tickets?
No. Your history is used only to answer your own routing requests. It is never used to train or improve models, and it is never shared with other customers or with outside LLM providers.
Are routing requests stored?
No. Incidents you send for routing are processed, answered and discarded. Only usage data such as token counts and cost is kept, for billing.
How do I sign up?
Sign up with a Google account or an email address. You get $1 of credit straight away, and you can reload from $5 by card when it runs out.
Can our AI agents call it?
Yes. The For agents page has the API, a tool definition you can give to Claude or ChatGPT, and the llms.txt files.
Does it pull tickets from my ITSM on its own?
Not yet. Today you upload history and call the API with new tickets. Direct connectors that read and update tickets are planned.