Parallel: several perspectives at once
Parallel distributes independent subtasks and then aggregates results.
What you will learn
- Explain: a workflow has explicit structure
- Apply the idea in an example: Parallel: several perspectives at once
- Recognize limitations and verify the exercise outcome
Parallel distributes independent subtasks and then aggregates results. Imagine three reviewers reading one description from different perspectives.
These steps refer to features identified in the application code. This material has not been validated on a live instance; check the documentation and the options in your version.
How it works, step by step
A workflow has explicit structure
A node is a stage, an edge is a transition, and state carries results between stages. A workflow has an application-defined path, although the model may choose some branches. Agentic systems include both workflows and dynamically acting agents. Nymrio offers ReAct, Reflection, Sequential, Orchestrator, Router, Parallel and Plan-and-Execute. Each changes coordination; it does not automatically make a model an expert. More agents mean more calls and contracts between roles. Start with the simplest solution that passes your tests.
Cost belongs to the whole path
Count model calls, input and output tokens, embeddings, reranking, vision and tools. Reasoning can consume billable tokens without an equally long visible response. Indexing costs and conversation costs differ. Use current prices rather than a permanent number from a course. A simple estimate is tokens/1,000,000 × price plus additional operations; application credits may have their own conversion. Compare cost per successful task, not just per call. Parallel execution can reduce latency while increasing total consumption. A budget needs a verified stopping condition.
Define success before optimizing
Prepare representative requests and expected answers or properties. Include ordinary cases, ambiguity, missing data and tool errors. For RAG, measure evidence retrieval separately from answer correctness. Recall@k measures the share of relevant evidence found in the first k results; precision@k measures the share of returned results that are relevant. For agents, also track actions, permissions and stopping. An LLM-as-judge can speed evaluation but needs calibration against humans. Do not change prompt, model and index simultaneously if you want to understand what improved results.
Detailed lab
Independence and aggregation
Splitter: “Distribute independent reviews of the same description”. Clarity: “Identify jargon and ambiguity”. Usefulness: “Check whether needs and limitations are explained”. Compatibility: “Compare claims against the catalog without assumptions”. Aggregator: “Retain sources, combine findings and explain disagreement”.
“LX-240 is perfect for any adapter” produces vagueness, overpromising and a contradiction with catalog adapter A-7. The final report recommends removing the generalization. Do not vote to determine physical compatibility. If a worker depends on another worker’s conclusion, move that stage into an appropriate chain.
The visual map
Follow the solid arrows for the main flow. Dashed blue arrows supply data or context; dashed pink arrows show feedback or returning results. Colors and shapes distinguish models, stores, decisions and outputs. Independent workers run concurrently; the aggregator combines their findings, not a majority vote. On smaller screens, scroll horizontally to follow the entire diagram.
A complete example
The splitter distributes clarity, usefulness and compatibility checks. Workers operate separately. The aggregator retains supported findings and explains disagreement.
splitter: TaskDistributor
worker_1: ClarityReviewer
worker_2: UsefulnessReviewer
worker_3: CompatibilityReviewer
aggregator: ReportWriter
Try it yourself
- Assign splitter, at least two workers and aggregator; independent branches receive the same input.
- Record the input, source and expected outcome before running the experiment. Use only the fictional data in the example.
- Follow the diagram stages. At every step record what information is received and produced; do not confuse intermediate output with the final outcome.
- Repeat after removing necessary information or making the input ambiguous. Check whether the system clarifies, stops or invents an answer.
- Compare with the explained solution. Keep the configuration, date, result and an explanation for differences. Change one thing and retest.
An explained solution
The final report distinguishes worker findings and does not treat majority voting as proof. A successful exercise lets you show the connection between input, stages and outcome. When information is missing, a cautious answer is more useful than invented details. Compare more than style: check conditions, sources and operations too.
When it helps and what can go wrong
A worker needing another worker’s output is not independent. Choose this approach when it improves a measured need. Keep a simple baseline and compare outcomes using identical inputs. One successful example does not establish reliability in every situation.
Check your understanding
Is multi-agent an eighth template?
No. It is an architectural category that can use several existing templates.
Are more simultaneous calls automatically cheaper?
No. Elapsed time and total consumption are different quantities.
What outcome should this exercise produce?
The final report distinguishes worker findings and does not treat majority voting as proof.
Words to remember
- Nod: A stage in an execution graph.
- Latență / Latency: Time until a useful result.
- Regresie / Regression: A change that breaks a previously correct case.
Sources and your next step
To prepare: W06 — Router: send requests to the right specialist