You’re not brief on instruments. Or fashions. Or frameworks.
What you’re brief on is a principled method to make use of them — at scale.
Constructing efficient generative AI workflows, particularly agentic ones, means navigating a combinatorial explosion of decisions.
Each new retriever, immediate technique, textual content splitter, embedding mannequin, or synthesizing LLM multiplies the area of doable workflows, leading to a search area with over 10²³ doable configurations.
Trial-and-error doesn’t scale. And model-level benchmarks don’t mirror how elements behave when stitched into full methods.
That’s why we constructed syftr — an open supply framework for mechanically figuring out Pareto-optimal workflows throughout accuracy, price, and latency constraints.
The complexity behind generative AI workflows
As an example how rapidly complexity compounds, think about even a comparatively easy RAG pipeline just like the one proven in Determine 1.
Every element—retriever, immediate technique, embedding mannequin, textual content splitter, synthesizing LLM—requires cautious choice and tuning. And past these selections, there’s an increasing panorama of end-to-end workflow methods, from single-agent workflows like ReAct and LATS to multi-agent workflows like CaptainAgent and Magentic-One.
What’s lacking is a scalable, principled solution to discover this configuration area.
That’s the place syftr is available in.
Its open supply framework makes use of multi-objective Bayesian Optimization to effectively seek for Pareto-optimal RAG workflows, balancing price, accuracy, and latency throughout configurations that will be unattainable to check manually.
Benchmarking Pareto-optimal workflows with syftr
As soon as syftr is utilized to a workflow configuration area, it surfaces candidate pipelines that obtain robust tradeoffs throughout key efficiency metrics.
The instance under exhibits syftr’s output on the CRAG (Complete RAG) Sports activities benchmark, highlighting workflows that keep excessive accuracy whereas considerably lowering price.
Whereas Determine 2 exhibits what syftr can ship, it’s equally vital to know how these outcomes are achieved.
On the core of syftr is a multi-objective search course of designed to effectively navigate huge workflow configuration areas. The framework prioritizes each efficiency and computational effectivity – important necessities for real-world experimentation at scale.
Since evaluating each workflow on this area isn’t possible, we usually consider round 500 workflows per run.
To make this course of much more environment friendly, syftr features a novel early stopping mechanism — Pareto Pruner — which halts analysis of workflows which might be unlikely to enhance the Pareto frontier. This considerably reduces computational price and search time whereas preserving outcome high quality.
Why present benchmarks aren’t sufficient
Whereas mannequin benchmarks, like MMLU, LiveBench, Chatbot Enviornment, and the Berkeley Operate-Calling Leaderboard, have superior our understanding of remoted mannequin capabilities, basis fashions hardly ever function alone in real-world manufacturing environments.
As a substitute, they’re usually one element — albeit a vital one — inside bigger, refined AI methods.
Measuring intrinsic mannequin efficiency is vital, however it leaves open vital system-level questions:
- How do you assemble a workflow that meets task-specific targets for accuracy, latency, and value?
- Which fashions do you have to use—and through which elements of the pipeline?
syftr addresses this hole by enabling automated, multi-objective analysis throughout complete workflows.
It captures nuanced tradeoffs that emerge solely when elements work together inside a broader pipeline, and systematically explores configuration areas which might be in any other case impractical to judge manually.
syftr is the primary open-source framework particularly designed to mechanically determine Pareto-optimal generative AI workflows that stability a number of competing goals concurrently — not simply accuracy, however latency and value as effectively.
It attracts inspiration from present analysis, together with:
- AutoRAG, which focuses solely on optimizing for accuracy
- Kapoor et al. ‘s work, AI Brokers That Matter, which emphasizes cost-controlled analysis to forestall incentivizing overly expensive, leaderboard-focused brokers. This precept serves as certainly one of our core analysis inspirations.
Importantly, syftr can be orthogonal to LLM-as-optimizer frameworks like Hint and TextGrad, and generic stream optimizers like DSPy. Such frameworks may be mixed with syftr to additional optimize prompts in workflows.
In early experiments, syftr first recognized Pareto-optimal workflows on the CRAG Sports activities benchmark.
We then utilized Hint to optimize prompts throughout all of these configurations — taking a two-stage method: multi-objective workflow search adopted by fine-grained immediate tuning.
The outcome: notable accuracy enhancements, particularly in low-cost workflows that originally exhibited decrease accuracy (these clustered within the lower-left of the Pareto frontier). These positive aspects counsel that post-hoc immediate optimization can meaningfully enhance efficiency, even in extremely cost-constrained settings.
This two-stage method — first multi-objective configuration search, then immediate refinement — highlights the advantages of mixing syftr with specialised downstream instruments, enabling modular and versatile workflow optimization methods.
Constructing and lengthening syftr’s search area
Syftr cleanly separates the workflow search area from the underlying optimization algorithm. This modular design allows customers to simply lengthen or customise the area, including or eradicating flows, fashions, and elements by enhancing configuration information.
The default implementation makes use of Multi-Goal Tree-of-Parzen-Estimators (MOTPE), however syftr helps swapping in different optimization methods.
Contributions of recent flows, modules, or algorithms are welcomed by way of pull request at github.com/datarobot/syftr.
Constructed on the shoulders of open supply
syftr builds on quite a lot of highly effective open supply libraries and frameworks:
- Ray for distributing and scaling search over massive clusters of CPUs and GPUs
- Ray Serve for autoscaling mannequin internet hosting
- Optuna for its versatile define-by-run interface (just like PyTorch’s keen execution) and help for state-of-the-art multi-objective optimization algorithms
- LlamaIndex for constructing refined agentic and non-agentic RAG workflows
- HuggingFace Datasets for quick, collaborative, and uniform dataset interface
- Hint for optimizing textual elements inside workflows, equivalent to prompts
syftr is framework-agnostic: workflows may be constructed utilizing any orchestration library or modeling stack. This flexibility permits customers to increase or adapt syftr to suit all kinds of tooling preferences.
Case examine: syftr on CRAG Sports activities
Benchmark setup
The CRAG benchmark dataset was launched by Meta for the KDD Cup 2024 and consists of three duties:
- Job 1: Retrieval summarization
- Job 2: Information graph and net retrieval
- Job 3: Finish-to-end RAG
syftr was evaluated on Job 3 (CRAG3), which incorporates 4,400 QA pairs spanning a variety of subjects. The official benchmark performs RAG over 50 webpages retrieved for every query.
To extend issue, we mixed all webpages throughout all questions right into a single corpus, making a extra reasonable, difficult retrieval setting.
Notice: Amazon Q pricing makes use of a per-user/month pricing mannequin, which differs from the per-query token-based price estimates used for syftr workflows.
Key observations and insights
Throughout datasets, syftr persistently surfaces significant optimization patterns:
- Non-agentic workflows dominate the Pareto frontier. They’re quicker and cheaper, main the optimizer to favor these configurations extra continuously than agentic ones.
- GPT-4o-mini continuously seems in Pareto-optimal flows, suggesting it affords a robust stability of high quality and value as a synthesizing LLM.
- Reasoning fashions like o3-mini carry out effectively on quantitative duties (e.g., FinanceBench, InfiniteBench), seemingly resulting from their multi-hop reasoning capabilities.
- Pareto frontiers finally flatten after an preliminary rise, with diminishing returns in accuracy relative to steep price will increase, underscoring the necessity for instruments like syftr that assist pinpoint environment friendly working factors.
We routinely discover that the workflow on the knee level of the Pareto frontier loses just some proportion factors in accuracy in comparison with essentially the most correct setup — whereas being 10x cheaper.
syftr makes it straightforward to seek out that candy spot.
Price of operating syftr
In our experiments, we allotted a funds of ~500 workflow evaluations per process. Though actual prices range primarily based on the dataset and search area complexity, we persistently recognized robust Pareto frontiers with a one-time search price of roughly $500 per use case.
We count on this price to lower as extra environment friendly search algorithms and area definitions are developed.
Importantly, this preliminary funding is minimal relative to the long-term positive aspects from deploying optimized workflows, whether or not by means of diminished compute utilization, improved accuracy, or higher consumer expertise in high-traffic methods.
For detailed outcomes throughout six benchmark duties, together with datasets past CRAG, discuss with the full syftr paper.
Getting began and contributing
To get began with syftr, clone or fork the repository on GitHub. Benchmark datasets can be found on HuggingFace, and syftr additionally helps user-defined datasets for customized experimentation.
The present search area consists of:
- 9 proprietary LLMs
- 11 embedding fashions
- 4 normal immediate methods
- 3 retrievers
- 4 textual content splitters (with parameter configurations)
- 4 agentic RAG flows and 1 non-agentic RAG stream, every with related hierarchical hyperparameters
New elements, equivalent to fashions, flows, or search modules, may be added or modified by way of configuration information. Detailed walkthroughs can be found to help customization.
syftr is developed absolutely within the open. We welcome contributions by way of pull requests, function proposals, and benchmark reviews. We’re significantly considering concepts that advance the analysis course or enhance the framework’s extensibility.
What’s forward for syftr
syftr remains to be evolving, with a number of energetic areas of analysis designed to increase its capabilities and sensible influence:
- Meta-learning
Presently, every search is carried out from scratch. We’re exploring meta-learning strategies that leverage prior runs throughout comparable duties to speed up and information future searches. - Multi-agent workflow analysis
Whereas multi-agent methods are gaining traction, they introduce extra complexity and value. We’re investigating how these workflows evaluate to single-agent and non-agentic pipelines, and when their tradeoffs are justified. - Composability with immediate optimization frameworks
syftr is complementary to instruments like DSPy, Hint, and TextGrad, which optimize textual elements inside workflows. We’re exploring methods to extra deeply combine these methods to collectively optimize construction and language. - Extra agentic duties
We began with question-answer duties, a vital manufacturing use case for brokers. Subsequent, we plan to quickly broaden syftr’s process repertoire to code technology, knowledge evaluation, and interpretation. We additionally invite the neighborhood to counsel extra duties for syftr to prioritize.
As these efforts progress, we goal to broaden syftr’s worth as a analysis device, a benchmarking framework, and a sensible assistant for system-level generative AI design.
For those who’re working on this area, we welcome your suggestions, concepts, and contributions.
Attempt the code, learn the analysis
To discover syftr additional, try the GitHub repository or learn the complete paper on ArXiv for particulars on methodology and outcomes.
Syftr has been accepted to seem on the Worldwide Convention on Automated Machine Studying (AutoML) in September, 2025 in New York Metropolis.
We look ahead to seeing what you construct and discovering what’s subsequent, collectively.
