Skip to content
Analyze a Decision →

The process, the computation library and the controls behind every report.

  1. Problem framing

    The problem, objective and constraints are restated; work continues only after confirmation.

  2. Sourced figures

    Public data and user-supplied documents, tied to a source and a version.

  3. Computational model

    A suitable mechanism is selected from the library and its assumptions are declared.

  4. Scenarios and stress tests

    Base, adverse and favourable cases are run, with sensitivity tested on the main variables.

  5. Checks and review

    Results are tested against defined checks; AI review runs in a separate context.

  6. Decision report

    Conclusion, breakpoint, conditions to change and sources, in a fixed structure.

Scale of the computation library

Computation mechanisms
99
Formulas
530
Worked-example checks
1,983
Historical stress scenarios
51
Public data source families
48
Reference classes
41
Business model templates
19
ISIC top-level sections covered
22

Commitments and data sources

  • Every figure is computed by the model

    Figures come from curated, controlled data, not from a model's own estimate.

  • Every finding carries its source

    Every number is tied to a run, a source or a stated assumption.

  • Your studies train no AI model

    A user's study content is not used to train general AI models.

  • Tight scenario control

    Scope, assumptions and price are confirmed before computing; no step runs outside the confirmed scope.

  • 11,000+

    project records

    Benchmark your plan against the cost overruns, schedule slippage and survival rates of past projects and firms.

  • 100+

    historical market shocks

    Test your cash flow against shocks drawn from observed market moves, including the 1997 and 2008 crises and COVID-19.

  • 99

    quantitative models

    Every calculation is made by financial and statistical algorithms, kept apart from a language model's guesswork.

  • 100%

    reproducible outputs

    Every result ties to its inputs, the model version and a source or a stated assumption, so you can check it.

Methods
  • Monte Carlo simulation
  • Stress tests from market history
  • Reverse stress test
  • Sensitivity analysis
  • Breakpoint search
  • Feasible region
  • Reference-class forecasting
  • Backtest
  • Linear programming
  • Markov chains
  • Chain-ladder
  • Erlang C queueing

99 quantitative models

Applies to the figures in a Scnro report.

Standards: UNIDO · HM Treasury · European Commission. Data: World Bank · IMF · OECD · ILO · Eurostat.

Prompts to copy are in Using Scnro.

Frequently asked questions

Points commonly raised before using Scnro.

Name and pronunciation

Scnro is pronounced “scenario”. The name reflects the function: simulating the scenarios of a decision.

AI assistant account requirements

Scnro connects to an AI assistant through an MCP connector, on Claude, ChatGPT and Gemini CLI. An assistant that allows custom connectors is required.

Difference from a general AI assistant

The same AI assistant is used, with a structured process added: problem framing, a calculation on a model, scenario runs and a source for every figure. The comparison table describes mechanisms; no measured comparison of accuracy is available.

Data sources

Public data from organisations such as the World Bank, IMF, OECD, ILO and Eurostat, together with documents supplied by the user. Each figure is tied to a source and a version.

Result checks

Each report carries a table of the automated checks that were run. Every plan can add an independent AI review. Both are advisory and do not replace review by a qualified person.

User data

Studies stay in the account; an assistant keeps its own chat. Once the Studio opens, it keeps a conversation 90 days after last use, via Anthropic. Personal data is not sold; study content is not used for training. Privacy Policy

Cost and Scnro Pennies

Signing up gives 10 free Scnro Pennies a month, no card. Pro and the one-off pack of 210 are not on sale yet. Inviting a friend gives each side 20; a first connection gives 30.

Scnro Web roadmap

Scnro Web, also called the Studio, is direct use in the browser on an Anthropic AI model that Scnro calls through its API. Not open yet, no release date; Claude, ChatGPT and Gemini CLI remain the channels.