AI agentIntake-to-pay

AI request routing agent

An AI request routing agent reads the structured request and compares it with your routing rules, for example which team handles software, which handles facilities and who owns a given budget.

  • People stay in control
  • Every action is logged
  • Start with a small sample
Classified request record
Routing map of owners by type
Leave and cover calendar
Vendor owner assignments
AI request routing agentPrepared for review
  • ✓Assigned owner or queue
  • ✓Reason for the route
  • ✓Notification to handler and requester
Handed toHandler receives the request with its context
5inputs it reads
6steps in the walkthrough
1point where a person decides
3outputs it prepares
See it run

Watch it work

Agent Record Person
Live walkthrough
  1. 1Classified requestRecord
  2. 2Read routing mapAgent
  3. 3Match ownerAgent
  4. 4Check workloadAgent
  5. 5Lead resolvesPerson
  6. 6Assign and notifyRecord

A labelled request reaches routing.

The agent reads who handles which type and size.

It picks the queue and the named handler.

It notes absence or a full queue.

A lead settles any request with no clear owner.

The handler is notified and the requester sees who has it.

The basics

What is an AI request routing agent?

It places the request in the correct queue and notifies the owner, adding a short note on why that destination was chosen.

Read the full explanation

When two rules conflict or none applies, it does not pick one. It parks the request for a procurement lead to direct.

Where it breaks

Why this step is hard by hand

Routing looks simple, but small errors send requests to the wrong desk, where they sit until someone notices.

A misrouted request waits in a queue nobody is watching, and the requester assumes it is moving.

Who handles what is often known only by a few people, so cover during absence is patchy.

A request can legitimately belong to IT and to finance, and nobody is sure who goes first.

Analysts forward requests by email, which leaves no record of why it went where it did.

When a team lead leaves, requests keep flowing to a person who is no longer there.

The path

How the agent does it

  1. 01Read the record

    The agent takes the category, amount, department and tags from the structured request.

  2. 02Match routing rules

    It looks for the rule that fits, such as software over a set value going to the IT sourcing queue.

  3. 03Check availability

    It checks the destination owner's status, including out-of-office and delegation settings.

  4. 04Assign and notify

    The request is placed in the queue, the owner is notified and the reason is logged.

  5. 05Park the exceptions

    Requests with no matching or with conflicting rules are held and flagged for a lead.

Want to see this flow on your own items?

Bring a small sample and watch Flo work through it.

Book a demo
Under the hood

Inside the work

AspectBy handWith the agent
OwnershipRequests sit in a shared inbox until someone claims them.Each request is offered to a named handler or queue on arrival.
Rule upkeepRouting knowledge lives in a few people's memory.Routing rules are written down and read each time.
AbsencesRequests wait for a handler who is on leave.Leave and cover are read from the routing map.
MisroutesA wrong team receives it and forwards it, losing days.The reason for each route is recorded so errors can be traced.
Requester viewRequesters do not know who is handling their request.The requester is told the handler's name or team.
People and control

Who decides, and what to watch

What stays with people

  • ✓
    Writing the routing rulesPeople decide who owns which kind of request and in what order.
  • ✓
    Resolving conflictsA lead settles cases where two teams could own the same request.
  • ✓
    Maintaining ownersProcurement keeps the list of queue owners and backups current.
  • ✓
    Overriding a routeAnyone with authority can redirect a request, and the change is recorded.

Mistakes to avoid

  • ×
    Rules that overlapTwo rules that both fire create exactly the ambiguity routing is meant to remove.
  • ×
    Stale owner listsRouting to someone who has moved teams sends requests to a dead end.
  • ×
    No reason recordedWithout a note, reviewers cannot tell whether a route was a rule or a mistake.
  • ×
    Routing before classificationSending a request on before it is properly labelled spreads the error forward.

Not sure where request routing fits in your process?

Talk it through with the team. We will map it to your steps and show where a person stays in control.

  • Works from a sample of your real items
  • Shows what Flo handles and what it hands over
  • No commitment to start
Rollout

Rolling it out

  1. 1Write the map

    Record who handles what for one department or class before the agent routes anything.

  2. 2Suggest only

    The agent proposes the owner and a person clicks to assign.

  3. 3Assign directly

    Allow direct assignment for rules that have never been overridden.

  4. 4Maintain the map

    Review misroutes each month and update owners after team changes.

Signs you would benefit

Requests wait unclaimedA shared inbox hides who is responsible.
Forwarding chains are commonThe same request is passed between three people.
Routing depends on one person's memoryCover during leave is improvised.
Requesters ask who has their requestNobody can answer quickly.
The numbers

How to measure it

Set your own baseline first. We do not publish benchmarks without a source you can check.
Metric 1

First-route accuracy

Accuracy (%) =
requests that reached the correct owner on the first routetotal requests routed× 100

It shows how often the destination was right without a redirect. A dip often follows an organisation change that the rules have not caught up with.

  • Count every manual redirect as a miss and review the pattern monthly.
  • Pair it with time spent in the queue before first action.
Questions

Frequently asked questions

Quick answers to what people ask most about request routing.

Book a demo
What does a request routing agent do?

It sends each purchase request to the right queue or owner according to your written rules, records why, and holds anything that fits no rule for a person to direct.

How is routing different from approval routing?

Request routing decides which team handles the request. Approval routing decides who must approve it. They use different rules, though the second often follows the first.

What if the right owner is away?

The agent can use the delegation or backup you have defined for that owner. If none is set, it flags the request to a lead rather than leaving it waiting.

Can people change a route?

Yes. Any authorised person can redirect a request. The change and the reason are logged so the rules can be improved.

Try request routing on your own data

Bring a small sample of real items. We will show what Flo handles on its own and where it hands over to your team.

  • People approve, sign and pay
  • Every action is logged
  • Start with a small sample