Geneva Forecast API · By RoadMap Technologies

One API call.
A forecast your P&L can trust.

Geneva evaluates statistical and ML forecasting models, picks the best fit per series automatically, and returns production-grade forecasts in under 500ms. The same engine behind Fortune 500 demand plans since 1988, now one POST away.

Get your free API key → Talk to an expert

Free tier · No credit card · First forecast in under 5 minutes

# pip install geneva-forecast
import geneva

forecast = geneva.forecast(
    series="sku_demand.csv",   # or JSON, or batch
    horizon=12,
    select="auto"             # automatic best-fit selection
)

print(forecast.mape)         # 8.4%
print(forecast.intervals)    # 80 / 95% confidence

Forecasting work since 1988 with teams at

Oracle · Dell · Samsung · Pfizer · Johnson & Johnson · Walmart · Amazon · Vertex

Why now

Bad forecasts are a line item. You just can’t see it yet.

The industry data is blunt: most planning stacks are still guessing, and the cost shows up as write-downs, stockouts, and missed quarters.

~$1T

Lost annually by global retailers to persistent stockouts.

Source: IHL Group, via Retail Dive

25-50%

Median forecast error (MAPE) across consumer goods: ~25% in food & beverage, ~50% in durables. Best-in-class hits ~20%.

Source: Gartner forecast accuracy benchmarks

70%

Of large organizations will adopt AI-based supply chain forecasting by 2030. The shift is happening now.

Source: Gartner, Sept 2025

88%

Of organizations use AI in at least one function, but 80%+ report no measurable EBIT impact yet. Tooling without workflow integration doesn’t move the P&L.

Source: McKinsey, State of AI 2025

47.9

ISM Manufacturing PMI (Dec 2025): contraction territory. Tight demand environments punish forecast error twice as hard.

Source: ISM PMI, Jan 2026

30-40%

Typical forecast-error reduction when teams replace manual or spreadsheet forecasting with Geneva’s best-fit engine.

Source: RoadMap customer engagements

These numbers describe the gap. Geneva closes it, without a 14-month platform rollout. It’s an API: deploy in hours, prove ROI in weeks.

Start free →

All third-party figures re-verified weekly on our Industry Stats page.

How it works

Your data to a defensible forecast in three steps.

Send your data

POST historical time series: sales, orders, pipeline, MRR, any metric. JSON, CSV, single series or thousands in one batch. No schema configuration.

Geneva picks the best model

The engine evaluates exponential smoothing, ARIMA, Prophet-class, gradient boosting, and neural forecasters, then selects the best fit per series, automatically.

Get forecasts instantly

Predictions with confidence intervals in under 500ms (P95). Consume via REST in your app, ERP, CRM, or dashboard, or via MCP in Claude and other AI agents.

Capabilities

Enterprise-grade forecasting. Developer-grade ergonomics.

🧠

Automatic best-fit selection

No model shopping. Geneva backtests statistical and ML models against your history and selects the winner per SKU, account, or series.

🔄

Batch forecasting

5,000 SKUs in one request. Parallel processing, one response, no job queues to babysit.

🤖

Native Claude MCP support

The only forecasting API with native Model Context Protocol support. Your AI agents call Geneva in natural language: forecast, analyze, surface risk.

⚡

Sub-500ms responses

Fast enough to render forecasts inline in your product UI. Real-time, not overnight batch jobs.

🔗

Works with your stack

Shopify, Salesforce, WooCommerce, Amazon FBA, HubSpot, Cin7, Google Sheets: pre-built paths, or plain REST for everything else.

📐

Confidence intervals included

Every forecast ships with 80/95% intervals and accuracy metrics, so finance and ops can argue about the plan, not the math.

Who uses Geneva

Four roles. One endpoint.

Demand Planner

Forecast 5,000 SKUs. Find $2.4M in working capital.

A CPG planner used Claude + Geneva to surface 187 SKUs with structural bias in under a minute.

See the workflow →

Finance Analyst

Reconcile pipeline vs. pattern.

FP&A surfaced $3.3M of Services optimism and $2.4M of Marketplace upside hiding inside an aligned-looking total.

See the workflow →

Developer

Ship “projected usage” in 50 lines.

One Geneva call per user, cached daily. Three product surfaces light up from the same forecast endpoint.

See the workflow →

Platform Lead

Forecast AI token usage before the budget breaks.

47.0B tokens forecast for the next 30 days, +22% growth, with the rate-limit ceiling break flagged 18 days in advance.

See the workflow →

Pricing

Start free. Scale when it’s earning.

No surprise bills. No annual lock-in to find out if it works.

Developer

$49/mo

✓ 100k forecasts/month

✓ REST API access

✓ Single & batch requests

✓ Email support

Start free trial

Business

$199/mo

✓ 1M forecasts/month

✓ Batch forecasting

✓ Claude MCP access

✓ Priority support

Start free trial

Enterprise

Custom

✓ Unlimited forecasts

✓ Dedicated infrastructure

✓ Custom integrations

✓ SLA + dedicated support

Contact sales

FAQ

Questions buyers actually ask.

How accurate are Geneva’s forecasts?

Geneva backtests its model library against your history and selects the best fit per series. Accuracy depends on data quality and history length, but teams replacing manual or spreadsheet forecasting typically cut forecast error 30-40%. For context, Gartner benchmarks median consumer-goods MAPE at 25-50%, so most teams have substantial room to recover.

How is this different from building with Prophet or ARIMA ourselves?

You can, and you’ll spend months on model selection, backtesting infrastructure, retraining pipelines, and scaling. Geneva is that infrastructure as a single API call, hardened across nearly four decades of Fortune 500 forecasting work. Your team ships the product; Geneva does the forecasting.

How does Claude MCP support work?

Geneva ships a native MCP server. Connect it to Claude (or any MCP-compatible agent) and your team asks for forecasts in plain language, like “forecast Q3 demand for the top 500 SKUs and flag structural bias,” and Geneva runs it. No custom code.

What data do I need to send?

Historical time series: dates and values. Sales, orders, pipeline, MRR, usage, any metric. JSON or CSV, single series or batch. No schema setup, no ETL project.

Is our data secure?

RoadMap Technologies is HITRUST risk-based 2-year certified and has handled regulated pharma and government forecasting data since 1988. Data is encrypted in transit and at rest.

How fast can we be live?

First forecast in under 5 minutes with the free tier. Production integration is typically hours to days. It’s a REST call, not a platform implementation.

The gap between your forecast and reality has a dollar figure.

Find out what it is. Free API key, no credit card, first forecast in five minutes.

Get started free →

Or book a 30-minute call and we’ll tell you whether Geneva fits, and how fast.

RoadMap Technologies, Inc · Beverly, MA · Est. 1988 · Industry Stats · Privacy · Contact

One API call.A forecast your P&L can trust.

Bad forecasts are a line item. You just can’t see it yet.

Your data to a defensible forecast in three steps.

Send your data

Geneva picks the best model

Get forecasts instantly

Enterprise-grade forecasting. Developer-grade ergonomics.

Automatic best-fit selection

Batch forecasting

Native Claude MCP support

Sub-500ms responses

Works with your stack

Confidence intervals included

Four roles. One endpoint.

Forecast 5,000 SKUs. Find $2.4M in working capital.

Reconcile pipeline vs. pattern.

Ship “projected usage” in 50 lines.

Forecast AI token usage before the budget breaks.

Start free. Scale when it’s earning.

Developer

Business

Enterprise

Questions buyers actually ask.

The gap between your forecast and reality has a dollar figure.

One API call.
A forecast your P&L can trust.