Verboisy gathers information from permitted web sources or prepares files you have access to. It checks results against your data contract and retains the references behind accepted records.
The goal, permissions and expected output are specified before collection.Every later step is tied back to those decisions.
Describe the question, fields, identity rules and allowed sources.
Research permitted web pages or parse explicitly granted files.
Check contracts, types and observed evidence before accepting records.
Export CSV or JSON alongside source references and saved run state.
Select a workflow to see what changes and what stays under your control.
Work inside specified domain boundaries.Define a schema, inspect accepted records and retain the material that supports their values.
field_1field_2Prepare declared files using deterministic adapters and explicit format selection.Local preparation does not require a model or a browser.
Call the synchronous or asynchronous API, or use the command line.Inspect result status, exports and the saved run manifest.
from verboisy import run
result = run(
goal="Your goal here",
schema="your_schema.json",
export="your_output.json"
)
print(result.status)Choose the interface that fits your workflow.The Desktop App, Python API and CLI use the same Verboisy engine, with guidance available in five languages.
Select Web research, Local preparation or Python integration using buttons, menus and guided steps. Set sources, files, exports, usage and spending options from clearly labelled controls.
Use Verboisy in your own Python programs with explicit goals, schemas, source boundaries and evidence-backed results.
Run jobs and inspect their state directly from the terminal, with repeatable commands and explicit control over data sources.
A conversational model is useful for exploring a question.A production data task also needs boundaries, structure and reviewable outputs.
This compares common workflows, not model accuracy or product benchmarks.
Verboisy is downloadable software that runs in your own environment, not a hosted service. Choose the licence that fits the people and workflows using it.
For independent professionals and early-stage teams starting with one user.
For organisations licensing Verboisy across several people and agreed environments.
Typical Team licences: $2,000–$10,000/year, depending on scope.
Discuss team licensingFor larger or specialised workflows that require a defined implementation agreement.
Custom engineering is quoted separately.
Discuss enterpriseVerboisy is available without a software licence fee for eligible students, self-taught learners, nonprofits and people in lower-income countries.
Students and self-taught learners studying data analysis, data science, machine learning or related subjects for educational purposes.
Nonprofits and community groups working in education, social welfare, humanitarian aid or environmental protection.
Individuals living in lower-income or economically underserved countries, considered on a case-by-case basis.
Free licences are reviewed individually.Eligibility, users, permitted use and duration are confirmed in a written agreement. Third-party model and API charges remain the applicant’s responsibility.
Request a free licenceProposed annual prices; licence terms and support are confirmed in writing. External API costs are separate. Source inspection does not permit production use.
Choose a guide in your language.Installation, CLI, Python API and Desktop App tutorials explain how to work with the same Verboisy engine.
Guides and tutorials are available in English, German, Italian, French and Spanish.
Verboisy is supplied to approved licensees as downloadable software with an activation key and guided installation instructions.
1. Download the authorized package sent with your licence and follow its operating-system setup guide.
2. Activate with the supplied key using the instructions delivered with the package. Keep the key private.
3. Check your installation and provider configuration before the first run.
Activation steps depend on the delivered build; an activation command is not published here.
verboisy --help
verboisy doctor
verboisy formats
verboisy model-configObtain the installation package and activation key through your licence. This is not a public download.
Describe a goal, list the columns you need, then run Verboisy from a terminal.
A new run requires --goal and --schema. Use --export to save CSV or JSON and --proof to produce a readable evidence report.
Restrict research with --include and --onlyinclude, exclude domains with --avoid, and control scope with --maxrows or --max-steps.
Inspect previous work with verboisy runs and verboisy status; continue an incomplete run with verboisy run --resume.
Replace the sample goal, field names, example.org and output path with your own permitted sources and requirements.
verboisy run \
--goal "Describe the information you need" \
--schema '[field_one,field_two]' \
--include example.org --onlyinclude \
--export output.json --proof
verboisy runs
verboisy run --resumeOnly research websites or process files you are authorized to access.
Integrate Verboisy into a Python program rather than launching a shell command.
Import run from verboisy and pass a goal, schema, and optional task, export and configuration. The sample below uses documented API arguments.
Read the returned status and manifest_path to inspect the outcome. For async applications, use await arun() rather than synchronous run() inside an event loop.
The API requires an authorized installed copy; model provider fees, where used, are separate.
from verboisy import run
from verboisy.utils.config import Settings
result = run(
goal="Describe your task here",
schema="schema.json",
task="task.json",
export="output.json",
config=Settings(sqlite_db_path="runs.db"),
)
print(result.status, result.manifest_path)Source-visible software. Running or evaluating it requires separate written permission.
Use the Verboisy Desktop App with buttons, selectable workflows and clear settings.Its interface, tutorials and instructions are available in multiple languages.
Web research: enter a research goal, choose output columns and allowed or excluded domains, then review export and proof settings.
Local preparation: pick an authorized file, select its format, set the schema or projection and choose where to export. Local stream processing does not call a model.
Python integration: open guided API examples and configuration instructions to connect Verboisy with your code.
Usage and spending: review provider settings, run limits and available cost information inside the Desktop App before starting. External model-provider billing and spending limits are managed separately.
Verboisy Desktop App
1. Select Web research, Local preparation or Python integration.
2. Choose sources or input files, required fields and export destination.
3. Review run limits and provider usage before starting.
4. Inspect results, evidence and saved runs.The Desktop App, Python API and CLI are all available interfaces. Use the language selector for tutorials and instructions.
Process files you have permission to use without first performing web research.
Use verboisy formats to inspect supported file formats and verboisy stream --help to see the large-file processing options.
For streaming, choose a granted --dataset-file and JSON schema or projection file. Use --rows-per-invocation to pause at a committed checkpoint; --resume continues a saved job.
The stream command handles local CSV, TSV, JSONL and text data without model calls; it does not perform automatic web discovery.
# Inspect supported formats and streaming options
verboisy formats
verboisy stream --help
# Check a saved stream job
verboisy stream-statusSource-visible software. Running or evaluating it requires separate written permission.
Specify fields using a simple list or JSON Schema Draft 2020-12 with local references.
Object schemas validate individual records. Array schemas also validate the final collection. Types are not silently coerced.
Evidence checks and schema validation are distinct from independent factual verification.
{
"type": "object",
"properties": {
"field_one": { "type": "string" },
"field_two": { "type": "string" }
},
"required": ["field_one"]
}Source-visible software. Running or evaluating it requires separate written permission.
Generate human-readable proof reports and retain references back to the observed material.
A proof report explains source references, accepted values and the checks used in a run.
Provenance does not prove that a third-party source is true or up to date.
verboisy run \
--goal "Describe the records to collect" \
--schema '[field_one,field_two]' \
--include example.org --onlyinclude \
--export output.json --proofSource-visible software. Running or evaluating it requires separate written permission.
Research workflows can use external model providers, with their charges billed separately.
Use verboisy model-config to inspect active model settings without provider calls or revealing credentials. Use verboisy doctor to check configuration.
Define source restrictions and run limits before starting research. In the Desktop App, review usage information and spending-related settings; the provider’s own billing safeguards still apply.
verboisy doctor
verboisy model-config
verboisy status --helpSource-visible software. Running or evaluating it requires separate written permission.
Use the separate stream interface for large, authorized local files.
It uses bounded buffers, a disk-backed ledger and durable row checkpoints. Supported types include CSV, TSV, JSONL, text and asset capture.
Streaming does not automatically add web discovery, semantic review, OCR or deduplication.
verboisy stream --helpSource-visible software. Running or evaluating it requires separate written permission.
Choose a declared parser for each supported input format.
CSV and TSV enforce consistent headers and rows. JSON and JSONL reject malformed objects. Text preserves lines. DOCX and XLSX have bounded extraction contracts.
Asset mode records bytes and checksums without interpreting their meaning.
verboisy formatsSource-visible software. Running or evaluating it requires separate written permission.
Approved customers receive licensed software, setup guidance and an activation key.
Individual, Team, Enterprise and eligible Community licences have different commercial terms. Contact us to request access or an evaluation.
The publicly visible source code does not itself grant permission to execute, modify, integrate or distribute Verboisy.
hello@verboisy.comSource-visible software. Running or evaluating it requires separate written permission.
Research and local preparation through three interfaces:Desktop App + Python API + CLI
Proprietary and source-visible.Inspection is not permission to run, test, integrate or redistribute.Written authorization is required.
This static site uses no accounts, forms, analytics scripts or tracking cookies. Hosting may process technical delivery logs. Email enquiries are handled by the site operator in Italy.
Report vulnerabilities privately with the affected version, impact and a minimal synthetic reproduction. This page does not authorize security testing or offer a bounty.
The Verboisy Source Inspection License 2.0 permits limited private review of authorized source material.It does not by itself grant rights to execute, benchmark, modify, deploy, redistribute or integrate the software.Those uses require separate written authorization, subject to applicable law and platform terms.
The full legally binding text is maintained with the source code. This website is a summary, not a replacement for that text.
Full authoritative license ↗Verboisy is operated by Emanuele Torrisi in Italy. This website has no contact form, login, advertising pixels, audience analytics or first-party tracking cookies. Switching language does not store a preference.
Netlify hosting may handle IP addresses and technical request information for site delivery and security. Emails sent to hello@verboisy.com are used to answer your enquiry. Where applicable, you may exercise GDPR rights through this address and complain to the Italian Garante or another competent authority.
Send a private report to hello@verboisy.com with the affected version or commit, expected security boundary, impact and a reproduction using synthetic data. Do not include secrets or customer data.
Testing is permitted only when you have lawful authorization and the necessary software rights. Reports do not imply a bounty, a response-time guarantee or permission to scan other systems.
Tell us which sources you intend to use, the output you need and any deployment constraints.We can discuss an evaluation, licensing or integration.
hello@verboisy.comCreated by Emanuele TorrisiResearch, validation and accountable data preparation.