Workflows quickstart - run a reusable research workflow

Workflows are in Beta. Start with a Valyu-curated workflow, inspect its inputs, and submit a research task. Completion time depends on the selected mode.

This example runs a company-profile workflow for NVIDIA. Workflows save the instructions and deliverables while you supply inputs such as the company name. To write your own research question instead, use the DeepResearch quickstart.

Get your API key and install an SDK

Create a key in the platform, check your balance and install an SDK.

pip install valyu        # valyu-py >= 2.12.0
export VALYU_API_KEY="your-api-key"

pnpm add valyu-js      # valyu-js >= 2.10.0
export VALYU_API_KEY="your-api-key"

List the Valyu workflows

Browse the available curated workflows and grab a slug.

from valyu import Valyu

valyu = Valyu()  # reads VALYU_API_KEY
listing = valyu.workflows.list(scope="valyu")
for wf in listing.workflows:
    print(wf.slug, wf.title, wf.vertical)

import { Valyu } from "valyu-js";

const valyu = new Valyu(process.env.VALYU_API_KEY!);
const listing = await valyu.workflows.list({ scope: "valyu" });
listing.workflows.forEach((wf) => console.log(wf.slug, wf.title, wf.vertical));

curl "https://api.valyu.ai/v1/workflows?scope=valyu" \
  -H "x-api-key: $VALYU_API_KEY"

Inspect a workflow's variables

Each workflow declares typed variables. Read them so you know exactly what workflow_params to send. The ib-company-profile workflow (“Company Profile”, investment-banking) takes one required variable, company.

profile = valyu.workflows.get("ib-company-profile")
print(profile.workflow.variables)

const profile = await valyu.workflows.get("ib-company-profile");
console.log(profile.workflow.variables);

curl "https://api.valyu.ai/v1/workflows/ib-company-profile" \
  -H "x-api-key: $VALYU_API_KEY"

Look at the variables array - each entry has a key, type (text / textarea / number / date / enum), and a required flag.

Run it

Pass workflow_id and workflow_params to deepresearch.create. Do not also send a freeform query - the two are mutually exclusive.

task = valyu.deepresearch.create(
    workflow_id="ib-company-profile",
    workflow_params={"company": "NVIDIA (NVDA)"},
)
print(task.deepresearch_id)
print(task.workflow)  # { "slug": "ib-company-profile", "version": 1 }

const task = await valyu.deepresearch.create({
  workflowId: "ib-company-profile",
  workflowParams: { company: "NVIDIA (NVDA)" },
});
console.log(task.deepresearch_id);
console.log(task.workflow); // { slug: "ib-company-profile", version: 1 }

curl -X POST "https://api.valyu.ai/v1/deepresearch/tasks" \
  -H "Content-Type: application/json" \
  -H "x-api-key: $VALYU_API_KEY" \
  -d '{
    "workflow_id": "ib-company-profile",
    "workflow_params": { "company": "NVIDIA (NVDA)" }
  }'

Poll for completion and get the deliverable

The run is a normal DeepResearch task. Poll until it finishes, then read the report and fetch deliverables.

result = valyu.deepresearch.wait(task.deepresearch_id)
print(result.status)
print(result.output)

const result = await valyu.deepresearch.wait(task.deepresearch_id);
console.log(result.status);
console.log(result.output);

curl "https://api.valyu.ai/v1/deepresearch/tasks/{task_id}/status" \
  -H "x-api-key: $VALYU_API_KEY"

Want to confirm the resolved prompt before paying for a run? Call valyu.workflows.preview(slug, workflow_params=...) (POST /v1/workflows/{slug}/preview) first - it resolves the workflow without running anything.

Start here

APIs

Finance workflows

Sciences workflows

Build and connect

Comparisons

Run a workflow

Next steps

Customize and version workflows

DeepResearch guide

​Next steps

Customize and version workflows

DeepResearch guide

Next steps