FlowBank: Query-Adaptive Agentic Workflows Optimization through Precompute-and-Reuse

📰 ArXiv cs.AI

Optimize LLM-based multi-agent systems with FlowBank, a query-adaptive approach that precomputes and reuses workflows to reduce inference costs and improve performance.

advanced Published 11 Jun 2026
Action Steps
  1. Analyze existing workflow optimization paradigms to identify trade-offs and limitations.
  2. Implement FlowBank's precompute-and-reuse approach to optimize workflows for LLM-based multi-agent systems.
  3. Evaluate the performance of FlowBank using metrics such as inference cost and workflow efficiency.
  4. Apply FlowBank to real-world applications, such as query-answering systems or decision-support systems.
  5. Compare the results of FlowBank with traditional task-level and query-level methods to demonstrate its effectiveness.
Who Needs to Know This

Machine learning engineers and researchers working on LLM-based multi-agent systems can benefit from FlowBank to optimize their workflows and improve overall system performance. This approach can be particularly useful for teams dealing with large-scale, complex workflows.

Key Insight

💡 FlowBank's precompute-and-reuse approach can significantly reduce inference costs and improve workflow efficiency in LLM-based multi-agent systems.

Share This
🚀 Optimize LLM-based multi-agent systems with FlowBank! 🤖

Full Article

Title: FlowBank: Query-Adaptive Agentic Workflows Optimization through Precompute-and-Reuse

Abstract:
arXiv:2606.11290v1 Announce Type: cross Abstract: Large Language Model (LLM)-based multi-agent systems are increasingly powerful, but current agentic workflow optimization paradigms make an unsatisfying trade-off. Task-level methods spend substantial offline compute yet deploy only a single workflow, leaving complementary candidates unused, while query-level methods synthesize a new workflow per query at substantial inference cost. Our motivating analysis shows these paradigms are more complemen
Read full paper → ← Back to Reads