You don’t have an AI problem. You have a decision problem.
Signal prices every decision on a twin of your P&L before anyone approves it, then measures what it caused against a holdout fixed in advance.
- Connect Your tables become a P&L twin.
- Price A range before approval.
- Approve With an owner and a reason.
- Measure Against a holdout fixed in advance.
- Competitor prices
- Price elasticity
- Unit costs
- Promo calendar
- Sell-out by region
- Demand forecast
- Stock by DC
- Freight and lead times
Pick a decision
Priced on the twin +R$1.1M to +R$1.9M margin in 12 weeks
Measured effect +R$1.46M
- With the decision
- Holdout, fixed in advance
- Priced range
Inside the priced range. Logged with its owner and the reason it was taken.
+R$1.1M to +R$1.9M margin in 12 weeks
+R$1.46M
Inside the priced range. Logged with its owner and the reason it was taken.
+R$0.6M to +R$1.2M margin in 12 weeks
+R$0.83M
Inside the priced range. The next promotion starts from this result.
+R$0.4M to +R$0.9M margin in 12 weeks
+R$0.31M
Below the priced range. The twin is recalibrated before the next call.
Three layers under every agent and module.
We’re a decision intelligence company. We build Signal, the system that recommends the trade-off, measures what it caused, and remembers why.
P&L Twin
A model of your P&L built from the tables you already have. It prices a decision before it is taken, and it is the shared context your agents and ours work from.
P&L Twin →Measurement
Each decision is registered with a control group before it starts, and only causal methods may promise an effect on the P&L.
Measurement →Decision orchestration
Every recommendation goes through independent approval and becomes an action with an owner and a due date, in a log that is never deleted.
Orchestration →Two ways to run on it: your agents, or our modules
Bring your own agents. You already run agents for pricing, planning or reporting. Signal gives them the twin’s context, and with our team their recommendations go through the same approval and measurement as any other decision. Early access.
/0.1RGM, a Pragma decision module. Price and price-pack architecture, promotion, portfolio and assortment, channel and commercial incentives.
/0.2Supply & Planning, a Pragma decision module. Forecast, inventory policy, service level, order to cash and the inputs to S&OP.
/0.3Either way, it starts with a Value Tree: 14 days to size the money on the table in that area, ending in a number with an owner and a date.
Start a Value TreeWhat the data showed, in four companies.
Contact
For pilots, partnerships, investors and hiring.
Message sent. We will reply to the e-mail you gave us.
hello@pragma.do