aboutsummaryrefslogtreecommitdiff
path: root/doc/workflow.md
diff options
context:
space:
mode:
authorOxbian <oxbian@mailbox.org>2025-03-04 21:15:42 -0500
committerOxbian <oxbian@mailbox.org>2025-03-04 21:15:42 -0500
commit43f26405e818aec791b28c50373843851fe1320e (patch)
tree81fe0cb2180afaebedc0edf65bd1c077ab267893 /doc/workflow.md
parentb9061a3e652cb7594397c38cd0078a47ddab960a (diff)
downloadNAI-43f26405e818aec791b28c50373843851fe1320e.tar.gz
NAI-43f26405e818aec791b28c50373843851fe1320e.zip
feat: routing request
Diffstat (limited to 'doc/workflow.md')
-rw-r--r--doc/workflow.md16
1 files changed, 16 insertions, 0 deletions
diff --git a/doc/workflow.md b/doc/workflow.md
new file mode 100644
index 0000000..04bfc07
--- /dev/null
+++ b/doc/workflow.md
@@ -0,0 +1,16 @@
+# Workflow
+---
+
+Néo AI use a system of workflow to give the user the impression of using an AI agents.
+
+
+**AI agents** are able to do actions automatically based on the user input, like searching the web, inside documents...
+Agents are designed to solve a specific problems and can be hard to make.
+
+**AI workflow** are able to follow a series of defined tasks, it's a defined pipeline to respond to the user request.
+Using the workflow system allows to built an AI that returns more powerful and trustable results. For example, in the case of searching factual information,
+like searching for personnalities or cultural events, the workflow can combine wikipédia results with LLM results, to limits LLM hallunication.
+
+## Coding workflow
+
+## Factual research workflow (Wikipédia)
ArKa projects. All rights to me, and your next child right arm.