Efficient determination of quantitative supply chain information

1 . A method in a computing system, comprising:

accessing an extract-transform-load (“ETL”) pipeline configured to act on dynamic supply chain data for a subject organization available from one or more data sources, the ETL pipeline comprising a plurality of tables and one or more ETL transform nodes, one or more first tables each being populated from the one or more data sources, one or more second tables each being populated by an ETL transform node of the one or more ETL transform nodes based on content of one or more feeder tables of the ETL transform node;

establishing in the ETL pipeline an insight ETL transform node identifying one or more of the plurality of tables as insight feeder tables, the insight ETL transform node specifying an insight mechanism, the insight ETL transform node being configured to invoke the specified insight mechanism passing contents of the insight feeder tables to populate an insight results data table among the plurality of tables; and

constructing an insight ontology object that references the insight results data table and contains a natural language text description of the insight results data table, the natural language text description being different from the properties of the insight results data table and describing the contents and significance of the insight results data table,

such that the insight results data table is configured to be automatically recurringly populated through the ETL pipeline, and is available for access and use by a language model configured to access the insight results data table based on the natural language text description contained in the constructed insight ontology object to service a prompt concerning dynamic supply chain data for the subject organization.

2 . The method of claim 1 wherein the insight mechanism specified by the insight ETL transform node is a procedural analytical algorithm.

3 . The method of claim 1 wherein the insight mechanism specified by the insight ETL transform node is a prompt for invoking a generative model.

4 . The method of claim 1 , further comprising:

adding the constructed insight ontology object to a set of resources available for use by the language model, the set of resources comprising:

a plurality of additional ontology objects, each of the additional ontology objects referencing a data object, and containing a natural language text description of the referenced data object describing its contents and significance, and

a natural language text description of the subject organization's supply chain institutional information comprising operational processes of the subject organization's supply chain.

5 . The method of claim 4 , further comprising:

receiving a prompt;

invoking the language model to process the prompt using the set of resources;

receiving a result from the invoked language model; and

causing the received result to be outputted.

6 . The method of claim 5 wherein the invoked language model is a specialized large language model trained to understand supply chain vocabulary, processes, and analysis.

7 . One or more instances of computer-readable media not constituting a signal per se, the one or more instance of computer-readable media collectively storing a data structure, the data structure comprising:

an extract-transform-load (“ETL”) pipeline configured to act on dynamic supply chain data for a subject organization, the ETL pipeline comprising:

a plurality of ETL transform nodes;

an insight ontology object that references a corresponding table configured to be populated by a distinguished ETL transform node and contains a natural language text description of a table configured to be populated by the distinguished ETL transform node describing its contents and significance, the natural language text description being distinct from the properties of the table; and

a multiplicity of tables, each of a plurality of first tables among the multiplicity of tables configured to be populated by an ETL transform node of the plurality of ETL transform nodes based upon content of one or more feeder tables of the ETL transform node, an insight ETL transform node of the plurality of ETL transform nodes specifying an insight mechanism to be invoked by the distinguished ETL transform node with contents of the distinguished ETL transform node's one or more feeder tables to populate a corresponding table with a derived insight result,

such that, when the distinguished ETL transform node operates within the ETL pipeline, the specified insight mechanism can be invoked to populate the corresponding table configured to be populated by the distinguished ETL transform node, and

such that, a language model is configured to access data stored in a table populated by an ETL transform node based on the natural language text description contained by the insight ontology object.

8 . The one or more instances of computer-readable media of claim 7 wherein the distinguished ETL transform node further specifies as a trigger condition a modification to at least one of the distinguished ETL transform node's feeder tables.

9 . The one or more instances of computer-readable media of claim 7 wherein the distinguished ETL transform node further specifies as a trigger condition an expiration of a designated refresh period.

10 . The one or more instances of computer-readable media of claim 7 wherein the insight mechanism specified by the insight ETL transform node is a procedural analytical algorithm.

11 . The one or more instances of computer-readable media of claim 7 wherein the insight mechanism specified by the insight ETL transform node is a prompt for invoking a generative model.

12 . One or more instances of computer-readable media not constituting a signal per se, the one or more instance of computer-readable media collectively having contents configured to cause a computing system to perform a method, the method comprising:

processing an extract-transform-load (“ETL”) pipeline configured to act on dynamic supply chain data for a subject organization available from one or more data sources, the ETL pipeline comprising a plurality of tables and one or more ETL transform nodes, one or more first tables each being populated from the one or more data sources, one or more second tables each being populated by an ETL transform node of the one or more ETL transform nodes based on content of one or more feeder tables of the ETL transform node;

as part of processing the ETL pipeline, triggering operation of an insight ETL transform node among the one or more ETL transform nodes, the insight ETL transform node identifying one or more of the plurality of tables as insight feeder tables, the insight ETL transform node specifying an insight mechanism; and

in response to triggering operation of the insight ETL transform node, invoking the specified insight mechanism passing contents of the insight feeder tables to populate an insight results table among the plurality of tables,

such that a language model is able to access the insight results table based on a natural language text description of an insight ontology object that references the insight results table and that is distinct from the properties of the insight results table.

13 . The one or more instances of computer-readable media of claim 12 , the method further comprising:

detecting at least one change to the insight feeder tables,

and wherein the triggering is performed in response to the detecting.

14 . The one or more instances of computer-readable media of claim 12 , the method further comprising:

detecting expiration of a refresh period of time,

and wherein the triggering is performed in response to the detecting.

15 . The one or more instances of computer-readable media of claim 12 , the method further comprising:

exposing the populated insight results table to a generative model to process a prompt concerning status of the subject organization's supply chain.

16 . The one or more instances of computer-readable media of claim 15 , the method further comprising:

receiving a prompt concerning dynamic supply chain data for the subject organization;

invoking the language model to process the prompt concerning dynamic supply chain data using a set of resources, the set of resources comprising an insight ontology object that references the insight results table and contains a natural language text description of the insight results table describing its contents and significance, and makes the insight results table available for access and use by the language model via the natural language description of the insight ontology object;

receiving a result from the language model; and

causing the received result to be outputted.