Agents
Tutorials for building agentic workflows and autonomous LLM-powered systems.
Autoresearch agent
Run an autonomous research loop that drives Claude Code in a GPU container to run experiments, then commits results and opens a pull request.
Parallelized autoresearch agent
Scale autoresearch with a code-mode MLE agent that batches train.py edits and runs sandbox experiments in parallel via flyte.map.
Code mode analytics agent
Chat with a dataset in the browser: Claude writes a Python program that runs in the Monty sandbox, with the heavy DuckDB query dispatched as a durable Flyte task.
AutoSec researcher agent
Fan out vulnerability analysis across C targets, hypothesize exploits with an LLM agent, and validate PoCs in an isolated sandbox.
Coding agent
Securely execute and iterate on LLM-generated code using a code agent with error reflection and retry logic.
Competitive intelligence agent
Fan out across competitors, extract source-cited market deltas with the You.com Search API, and build a knowledge-graph-ready intelligence table.
Compliance monitoring agent
Monitor trusted regulatory sources with the You.com Research API and route citation-precise findings to the right team.
Deep research
Build an agentic workflow for deep research with multi-step reasoning and evaluation.
LangGraph research agent
Combine LangGraph control flow with Flyte tasks for multi-topic web research with quality-check loops.
Field data enrichment agent
Enrich geo-tagged operational events with real-world public context using the You.com Search API with country and freshness targeting.
MLE Bot: autonomous ML engineer
An autonomous ML agent that designs, runs, and iterates on experiments using Flyte’s durable sandbox for safe LLM-generated code execution.
Support resolution agent
Ground support tickets in fresh public sources via the You.com Research API and draft cited, customer-ready replies for human review.