# US 2024/0176958 A1 · Prompting language models with workflow plans > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2024/0176958 A1 - Document ID: US20240176958A1 - Status: Published application - Kind: A1 - Application number: 18/072,479 - Filed: 2022-11-30 - Published: 2024-05-30 - Earliest claimed: 2022-11-30 - Assignees: ServiceNow, Inc. - Applicants: ServiceNow, Inc. - Inventors: Stefania Evelina Raimondo; Christopher Joseph Pal; Hector Luis Palacios Verdes; David María Vázquez Bermúdez; Xiaotian Liu - Web page: https://patentgrep.com/patent/US20240176958A1 ## CPC - G06F 40/40: Processing or translation of natural language (natural language analysis G06F40/20; semantic analysis G06F40/30) - G06F 40/35: Discourse or dialogue representation - G06N 3/006: based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO] - G06N 3/044: Recurrent networks, e.g. Hopfield networks - G06N 3/045: Combinations of networks - G06N 3/0455: Auto-encoder networks; Encoder-decoder networks - G06N 3/08: Learning methods - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract A pre-trained natural language processing machine learning model is received. The pre-trained natural language processing machine learning model is tuned. A workflow to be implemented by a chatbot to complete a task-oriented dialog is received. A representation of one or more plan steps based on a current progression state of the workflow is dynamically determined. At least the representation of the one or more plan steps is provided as an input to the pre-trained natural language processing machine learning model to guide the pre-trained natural language processing machine learning model. An output action and/or an utterance for the task-oriented dialog is received from the pre-trained natural language processing machine learning model. ## Claims ### Claim 1 A method, comprising: receiving a pre-trained natural language processing machine learning model; receiving workflow data indicating a workflow for a dialog with a chatbot associated with performing a task; determining a representation of one or more plan steps based on a current progression state of the workflow; providing additional input data to the pre-trained natural language processing machine learning model including at least the representation of the one or more plan steps; and determining, via the pre-trained natural language processing machine learning model, an output action or an utterance associated with the dialog. ### Claim 2 (depends on 1) The method of claim 1, wherein the pre-trained natural language processing machine learning model is configured to receive text data and output text data. ### Claim 3 (depends on 1) The method of claim 1, wherein the pre-trained natural language processing machine is learning model includes an encoder-decoder structure. ### Claim 4 (depends on 1) The method of claim 1, further comprising tuning the pre-trained natural language processing machine learning model. ### Claim 5 (depends on 4) The method of claim 4, wherein tuning the pre-trained natural language processing machine learning model includes further training the pre-trained natural language processing machine learning model using labeled action plans comprised of actions to complete multi-step tasks. ### Claim 6 (depends on 1) The method of claim 1, wherein the workflow includes a sequence of actions required to perform a task correctly. ### Claim 7 (depends on 1) The method of claim 1, wherein the chatbot is configured to transmit user utterances to the pre-trained natural language processing machine learning model. ### Claim 8 (depends on 1) The method of claim 1, wherein the chatbot is configured to transmit the output action or the utterance from the pre-trained natural language processing machine learning model to a dialog partner of the chatbot. ### Claim 9 (depends on 1) The method of claim 1, wherein the representation of the one or more plan steps includes natural language text describing specified actions. ### Claim 10 (depends on 1) The method of claim 1, wherein the representation of the one or more plan steps includes text generated using a specified syntax. ### Claim 11 (depends on 11) The method of claim 11, wherein the specified syntax is associated with a Planning Domain Definition Language. ### Claim 12 (depends on 1) The method of claim 1, wherein the workflow is associated with a sequence of application programming interface calls. ### Claim 13 (depends on 1) The method of claim 1, further comprising providing a representation of a dialog history of prior utterances in the dialog to the pre-trained natural language processing machine learning model. ### Claim 14 (depends on 13) The method of claim 13, wherein the representation of the dialog history includes a concatenation of string representations of the prior utterances in the dialog. ### Claim 15 (depends on 1) The method of claim 1, wherein the workflow is selected from a list of valid workflows. ### Claim 16 (depends on 1) The method of claim 1, wherein the task is an information technology task. ### Claim 17 (depends on 1) The method of claim 1, wherein the pre-trained natural language processing machine learning model has been trained on data that includes an action-based conversations dataset. ### Claim 18 (depends on 1) The method of claim 1, wherein completing the dialog has been formulated as an automated planner problem. ### Claim 19 A system, comprising: one or more processors configured to: receive a pre-trained natural language processing machine learning model; receive workflow data indicating a workflow for a dialog with a chatbot associated with performing a task; determine a representation of one or more plan steps based on a current progression state of the workflow; provide additional input data to the pre-trained natural language processing machine learning model including at least the representation of the one or more plan steps; and determine, via the pre-trained natural language processing machine learning model, an output action or an utterance associated with the dialog; and a memory coupled to at least one of the one or more processors and configured to provide at least one of the one or more processors with instructions. ### Claim 20 A computer program product embodied in a non-transitory computer readable medium and comprising computer instructions for: receiving a pre-trained natural language processing machine learning model; receiving workflow data indicating a workflow for a dialog with a chatbot associated with performing a task; determining a representation of one or more plan steps based on a current progression state of the workflow; providing additional input data to the pre-trained natural language processing machine learning model including at least the representation of the one or more plan steps; and determining, via the pre-trained natural language processing machine learning model, an output action or an utterance associated with the dialog.