Routespring Automation
Measured against your team before they run alone. Paid on results.
We take one workflow your team already runs, with its documents, systems and written rules, and deploy an agent that finishes the routine cases end to end. A person keeps every decision the policy reserves for people. Nothing runs alone until it beats your own team on your own files.
Shadow mode · week 4
Team and agent, side by sideTest set closed · Corrections join it weekly
Each one is a desk that already exists, with people doing the handoffs. Three recover money. Three remove work. All six have a number that says whether it worked.
Chase, read and itemize every crew hotel folio, check it against the contract, and raise the dispute.
For airline finance, crew travel desks, tmcs.
See the workflowMatch every invoiced room night to a crew booking and the contract, and recover what was overbilled.
For airline finance and procurement.
See the workflowTrack every unused ticket, credit and voucher, and apply it before it expires.
For airline staff travel, corporate travel managers, tmcs.
See the workflowClear airline schedule changes from the queue: classify, protect, notify, reissue.
For tmcs, airline crew travel desks, staff travel.
See the workflowFor airlines: find violations and issue the memo. For agencies: build the dispute and win it.
For airline revenue accounting, tmcs.
See the workflowDecide EU261, UK261, APPR and US refund claims under your written policy, and pay within limits.
For airline customer relations and legal.
See the workflowThe agent reads from your systems and never writes to them directly. Policy lives in code, judgement in the model, and every result comes back as a batch a person can undo.
01 · Your systems
Read only. Nothing here is changed directly.
02 · Agent runtime
Policy in code
Eligibility, limits, authority and the audit trail. Deterministic, versioned, tested.
Model
Reads the evidence, weighs conflicts, applies the written policy, writes down its reasoning.
Test set
Your 50 graded files plus every correction since. Run before each policy change.
Runs in your tenant or ours. ISO/IEC 27001 controls. No training on your data.
03 · What comes back
Batch write-back
Validated rows, stamped with the policy version, one-click undo.
Exception to a person
The facts gathered, the reasoning attached, the decision left to your team.
Error watch
Every row checked for eight weeks, then a sample. Weekly report to the owner.
The model is rarely the problem. Rollouts stall on four things, and each one has a rule.
Failure 1: No business owner
One person who runs the workflow today owns the agent after launch, with the authority to change the policy it follows.
Failure 2: No definition of success
Before any build, two of your people grade the same 50 closed files blind. Where they differ is the human baseline and the agent’s miss budget.
Failure 3: No guardrails
Code keeps eligibility, limits, authority and the audit trail. The model reads the evidence, weighs conflicts, applies the written policy and writes down its reasoning.
Failure 4: Change management as an afterthought
The people whose work changes grade the test set, watch shadow mode and correct the agent. Their corrections become the test set.
Week 1
One workflow your team already runs, with its documents, systems and written rules, one owner on your side, and one number that decides whether it worked.
Weeks 1 to 2
Fifty closed files with known outcomes, graded blind by two of your people. We write the agent without looking at them, run it, and report false positives and false negatives against the human baseline.
Weeks 3 to 8
About 100 new files run beside your team with no answers given. Your team compares weekly. The test set stays closed while we tune, so the agent is never fitted to it.
Weeks 9 to 10
The agent finishes the routine cases alone. Every batch is stamped with the policy version and can be undone. Exceptions go to a person with the facts gathered.
After go-live
Every finished case is checked for eight weeks, then a sample. Every correction joins the test set. The error rate is reported to the owner weekly.
| An internal team | A consulting firm | A weekend prototype | Routespring Automation | |
|---|---|---|---|---|
| Starts from | A backlog and a hiring plan | A discovery phase | A demo on sample data | Your 50 closed files |
| Definition of correct | Written later, if at all | Written in the readout | None | Before any build |
| Measured against people | Rarely | Sometimes | No | Always, blind |
| Runs the same desk | No | No | No | Yes, daily |
| Paid on | Headcount | Time | Nothing | Results |
Where a workflow touches Routespring bookings, these are the contracts your integration team will use. Where it lives in your own systems, the agent reads and writes there and none of this is required.
The 50-file test
Send 50 closed files under NDA. Within two weeks you get the agent’s error rate against your own people and, for recovery workflows, the dollars it would have found.
10-week pilot
Covers setup, the 50-file test write-up, shadow mode beside your team and a go or no-go report with the measured numbers.
Production, recovery workflows
Invoice audit, unused tickets and debit memos. A monthly minimum of $3,000. No platform fee.
Production, processing workflows
Reconciliation, schedule changes and claims. Priced per file handled, with a monthly minimum of $3,000. No platform fee.
Cancel on 30 days’ notice after the first eight production weeks. Your data is never used to train a model.
Routespring runs crew travel for airlines: hotels, disruptions, payments and the reconciliation that follows. The automation team is the people who built that platform and the operations team that works the desk every day. The method on this page came from automating our own work first.
The four rules and the 50-file method were presented to technology executives at an OpenAI event in San Francisco. The same material is on this site, in full, so you can hold us to it.
Events and conversationsOctober 22, 2026
OpenAI, San Francisco
Tarun Upaday, Routespring’s CEO, presented at an OpenAI event for technology leaders in San Francisco. The session covered why agent rollouts stall in operations: no business owner, no definition of correct, no guardrails, and change management left to the end. The four rules on this page are the answer.
Read the four rulesBuild versus buy
When an internal team should build the automation and when it should not.
Running an automation pilot
What a pilot needs from operations, finance and IT.
Integration readiness
The data and access questions to settle before shadow mode.
Crew travel RFP template
If the work goes to tender, the questions worth asking every vendor.
Fifty closed cases from the workflow you want to change, under NDA. Within two weeks you get the agent’s error rate against your own team and, for recovery workflows, the dollars it would have found. No charge.
