Checked against the rules, and the recommendation stands.
Intelligence décisionnelle pour les équipes opérationnelles
Décrivez le problème. Obtenez la décision.
Transformez un problème opérationnel en une décision défendable — avec l'impact, le niveau de confiance, les contraintes et les justificatifs.
Saisie en langage courant · recommandations vérifiées · interface EN/FR/ES/DE
- Input
- Plain language
- Checked
- Against your rules
- Output
- Recommendation + report
A decision, end to end
The brief, the model it built, and the answer it checked.
Three worked examples, each solved and verified ahead of time. Change the constraint and watch which part of the answer moves.
Brief
One van, five drop-offs, back to the depot. The stops sit on a city grid — A(2,4), B(5,1), C(6,5), D(1,7), E(8,3), depot at (0,0) — and driving distance is the grid distance in km. Which order is shortest?
Written the way someone would say it. The quantities marked above are what Mongeflow reads out of it.
The model
- Goal
- Shortest total driving distance for one van, returning to the depot
- Stops
- A(2,4) B(5,1) C(6,5) D(1,7) E(8,3) from the brief
- Depot
- (0,0) from the brief
- Distance
- grid distance in km from the brief
Constraints
- Every stop visited exactly once
- The round returns to the depot
Stops on the round
How many stops are on the round.
Both options were solved and checked ahead of time. Changing this switches between them — it does not run a new solve in your browser.
Decision
- Reading
- Checking
- Computing
- Verifying
- Ready
Round distance32 km
The shortest round possible — all 120 stop orders were measured, and none beats 32 km.
The plan, and its arithmetic
- Drive Depot → A → D → C → E → B → Depot
- Legs: 6 + 4 + 7 + 4 + 5 + 6 = 32 km
- Longest leg: D → C at 7 km
Compared with
Driving the stops in list order (A, B, C, D, E): 46 km — 14 km longer.
Checks
- Passed: Every stop visited exactly once
- Passed: Round returns to the depot
- Passed: All 120 stop orders measured
Add a sixth stop at (4,8) and the best round grows only 2 km, to 34 km.
The product
Watch the decision take shape
Describe the problem in business language and follow the answer as it forms — the recommendation assembling, every business rule checked, the result ready to review.
The experience
First-class decision experience
Four steps carry a written brief to a shared report, and you see each one happen.
01 · Describe
A brief in business language
Write the problem the way you would say it. Mongeflow shows back the goal, the rules, and any missing details before it recommends anything. No black box.
02 · Decide
A recommendation, checked
Expected impact, confidence, and tradeoffs, stated plainly — every result checked against your rules before it is shown.
03 · Explore
What-ifs, precomputed
See how the answer moves when the inputs do — alternatives and tradeoffs you can compare before you commit.
04 · Share
From decision to report
Approve it, adjust it, or send a stakeholder-ready report or data export. The API reads your decisions and signs every webhook.
Six domains
Delivery rounds, rosters, budgets, forecasts
Simulation
Probability forecasting, discrete-event and queueing models, sequential decisions under uncertainty.
Learn moreTry the match
Say it in your words
The match appears here — drawn from the shipped catalog, not a live solve.
The difference
Hand-built models vs the solved state
| Capability | Hand-built | Mongeflow |
|---|---|---|
| The understood problem, shown back and editable | ||
| Missing details named before recommending | ||
| A check on every recommendation | ||
| Marked verified only when all checks pass | ||
| Report and data exports | ||
| API reads + signed webhooks | ||
| Weeks of manual analysis |
Verified daily
Checked this morning. And every morning.
A verification suite runs Mongeflow's answer classes against the live product every morning — profit plans, rosters, routes, forecasts — and each answer is checked before the day is called green.
The API
One request. One answer.
Decisions are made in the product. The API reads what you decided, pulls a supported report or data format, and signs every webhook it sends.
# Read what was solved
curl https://mongeflow.com/api/v1/decisions?status=solved \
-H "Authorization: Bearer mf_live_xxxxxxxxxxxxxxxxx"
# → [{ "id": "…", "title": "…", "status": "solved",
# "savings_total": …, "decision_type_label": "Routing" }]
# Pull the deliverable
curl -X POST https://mongeflow.com/api/v1/exports \
-H "Authorization: Bearer mf_live_xxxxxxxxxxxxxxxxx" \
-d '{ "analysis_id": "…", "format": "pdf" }'
Weeks of analysis become one description.
The structuring, the tradeoffs, and the recommendation that once needed a dedicated analyst — returned from a single written description.