Third-party service for suggesting resources for a received message

A system, comprising:

• at least one server computer comprising at least one processor and at least one memory, the at least one server computer configured to:
• receive, via a first API request from a first entity, a first message of a first conversation, wherein the first conversation includes a first user;
• compute a first conversation feature vector by processing the first message with a first neural network;
• select a first suggested resource corresponding to the first message using the first conversation feature vector;
• return a first API response, wherein the first API response includes the first suggested resource;
• receive, via a second API request from a second entity, a second message of a second conversation, wherein the second conversation includes a second user;
• compute a second conversation feature vector by processing the second message with the first neural network;
• determine a second suggested resource corresponding to the second message using the second conversation feature vector;
• and
• return a second API response, wherein the second API response includes the second suggested resource.

One or more non-transitory computer-readable media comprising computer executable instructions that, when executed, cause at least one processor to perform actions comprising:

• receiving, via a first API request from a first entity, a first message of a first conversation, wherein the first conversation includes a first user;
• computing a first conversation feature vector by processing the first message with a first neural network;
• determining a first suggested resource corresponding to the first message using the first conversation feature vector;
• returning a first API response, wherein the first API response includes the first suggested resource;
• receiving, via a second API request from a second entity, a second message of a second conversation, wherein the second conversation includes a second user;
• computing a second conversation feature vector by processing the second message with the first neural network;
• selecting a second suggested resource corresponding to the second message using the second conversation feature vector;
• and
• returning a second API response, wherein the second API response includes the second suggested resource.