Negotiator Agent
Negotiates tail spend, order by order
Limit
Negotiates within the agreed range and approval limit; anything outside it goes to the buyer with the counteroffer ready.
UpFlux + TOTVS Datasul
AI in procurement inside TOTVS Datasul.
Process mining and AI agents running inside Datasul, from purchase request to purchase order, respecting the ERP's own approval limits and approvals. The buyer stays in the environment they already use; quotations for the tail stop waiting.
Integration through the Datasul APIs, available on the TOTVS StoreA Major Contender in the Everest Group PEAK Matrix® 2026
See also: UpFlux + TOTVS · TOTVS Protheus
// digital teamrunning
TOTVS Datasul · Supply
- Request 31877 · PPENegotiator Agent
- Order 55120 · valvesprice 18% above reference
- Order 55093 · packagingdelivery 2 days late
- Orders 55130 and 55131 · same itemapproval skipped
Every order follows Datasul's approval flow and is recorded in the ERP itself.
// digital team
Who works in your Datasul
Each agent has a role, a limit and an audit trail. Anything past the limit goes back to a person, with the case ready.
Follow-up Agent
Chases suppliers, sends reminders and keeps the process moving
Limit
Follows up on its own in the agreed tone; critical delays go to whoever owns the delivery.
Prioritizer
Orders the queue by impact in reais
Limit
Orders the queue; never removes items or changes owners.
Status Agent
Answers “where is my order?”
Limit
Informs; never changes the case or the due date.
Price Anomaly Detector
Catches prices paid above the reference
Limit
Flags; renegotiating belongs to the buyer or the Negotiator Agent.
Approval Limit Auditor
Watches who approves what
Limit
Flags the case; fixing it belongs to the controls team.
// what runs in your Datasul
Process mining, agents and RoAI, native in Datasul.
Process Intelligence over supply management
UpFlux reads the request, quotation, purchase order and receiving events from Datasul and rebuilds the real process: quotations left open, orders without a quotation, approvals that wait and rework per establishment.
Negotiator Agent writing to Datasul
The agent takes the approved request, checks the price history in Datasul itself, quotes and negotiates with the supplier within the approval limit and writes the purchase order through the API, with an audit trail per decision.
Nous routine agents
Expediting, supplier chasing and delivery status are executed by the agents inside the Datasul flow. The buyer decides by exception.
RoAI measuring the return in hard currency
Every agent action stays auditable in RoAI, with savings measured against the best price already paid in your base. The CFO sees, in hard currency, what AI returned in Datasul.
// rollout
Diagnostic in 2 weeks, with no migration project.
UpFlux does not ask for Datasul customization. You export the supply events, and the first map of the real process arrives ready, with the target in hard currency for what is addressable.
01
Native connection
Integration through the standard Datasul APIs, installed from the TOTVS Store. No parallel platform, no migration.
02
Diagnostic in 2 weeks
From the event export, we deliver the map of the real process and the target in hard currency: variants, bottlenecks and addressable return.
03
Agents in production
The first agent goes live within the approval flow that already exists, knowing where to act and how the result will be measured in RoAI.
// frequently asked
Frequently asked
Questions from Datasul teams evaluating process mining and AI agents in supply management.
Is there automatic AI quotation in Datasul?
Yes. UpFlux runs an AI agent inside TOTVS Datasul, available on the TOTVS Store, that takes the approved request, quotes and negotiates with suppliers against the price history in the ERP itself and writes the purchase order through the API, respecting Datasul's approvals and approval limits.
What does the UpFlux agent do that native Datasul quotation does not?
Native quotation organizes proposals; someone still has to request, chase, compare and negotiate. The UpFlux agent runs that cycle for tail items, within the approval limit, and returns the finished purchase order to Datasul. The buyer approves only the exception, and every decision stays auditable in RoAI.
How do you find where purchasing gets stuck in Datasul?
With process mining on the request, quotation, purchase order and receiving events. UpFlux rebuilds the real process per establishment and shows open quotations, orders without a quotation and approvals that wait, in two weeks from the event export, without customizing the ERP.
Does UpFlux need an integration project in Datasul?
No. The integration uses the standard Datasul APIs and is installed from the TOTVS Store. There is no data migration and no parallel platform; the buyer stays in the environment they already use.
Want to see UpFlux running in your Datasul?
In two weeks, we deliver the map of the real process and the target in hard currency, straight from the Datasul supply events.
Talk to UpFlux