back_office_ops · saas · workflow
ART·E: Building an Email Research Agent with Reinforcement Learning That Beats o3
Searching an email inbox to answer natural-language questions required manually opening the inbox, devising keywords, and reading through search results — a process the author wanted AI to replace.
How it works
Common implementation structure
How this type of workflow is generally built, generalized across documented cases — not tied to any one vendor's stack. Click any stage to read what happens there. Specific products that implement these stages appear in “Tools commonly seen” below.
Stage 1 · User submits email query
The agent is given an initial prompt defining the task: answer the user's question based on the contents of their inbox.
Tools used
ARTGRPOsqliteFTS5vLLMUnslothSkypilotRunpodWeights & Biases
Outcome
ART·E is faster, cheaper, and more accurate than o3 on the email search task, taking almost a full turn less per query while hallucinating fewer wrong answers. Training cost approximately $80 and completed in just under a day.
Results
Time savedjust under a day
Volumefaster, cheaper, and more accurate than o3
Cost replacedabout $80
Grounding & classification
Source type: technical build writeup
29 fields verified against source quotes, 1 dropped as unverifiable.
agentic workflowai agententerprise searchemailbuilder submittedfailure mode describedmetric backedtools describedworkflow describedsoftwareaccuracy improvementcost reductioncycle time reductionerror reductiontechnical build writeupback office opsagentic task executionrag answering