Conversation import
Import structured chatbot or support conversations and turn customer messages into reviewable samples.
IntentOps Studio 1.0.0 is available now
Turn customer conversations into local intent intelligence.
IntentOps Studio imports support conversations, extracts reviewable samples, builds intent datasets, trains local ML.NET classifiers, evaluates model quality, generates recommendations, and exports reports.
Offline-first
Encrypted projects
Local ML.NET training
Reports
An offline workspace for practical intent operations.
Built for teams that need to understand customer requests, prepare reviewed intent datasets, train local classifiers, and document model quality without sending data to a cloud service.
Human review
Confirm, correct, mark unknown, or discard samples before they become training data.
Dataset builder
Create intent datasets with minimum examples, weak-intent filtering, class balancing, and export packages.
Local model loop
Train local ML.NET classifiers, evaluate metrics, identify gaps, and generate improvement recommendations.
Guide every project from import to reports.
The project dashboard shows the whole intent-operations pipeline and helps users move through each step: business context, conversations, samples, datasets, models, evaluation, recommendations, and reports.
Create encrypted or unencrypted local projects.
Track project readiness from one dashboard.
Keep imported conversations and training artifacts on the user's machine.
Open the next best workflow step without guessing where to go next.
From raw conversations to reviewed training data.
IntentOps Studio keeps humans in the loop so noisy messages do not become unreliable model examples.
Raw chatbot exports
- Contain greetings, order identifiers, emails, status replies, and low-value noise.
- May mix multiple intents inside the same conversation.
- Are hard to use directly as reliable training data.
- Do not explain which intents are weak or underrepresented.
- Often require manual spreadsheet cleanup before training.
IntentOps workflow
- Extracts useful customer messages into reviewable samples.
- Lets reviewers confirm, correct, mark unknown, or discard samples.
- Builds structured intent datasets with metadata and notes.
- Trains and evaluates local models inside the same project.
- Generates recommendations and reports for the next improvement cycle.
Review samples before they become a dataset.
Sample review is where intent quality starts. Confirm good suggestions, correct wrong labels, mark unclear messages as unknown, and discard examples that would pollute the dataset.
Filter samples by status, source, confidence, and intent.
Correct the confirmed intent without leaving the review screen.
Keep only meaningful customer messages for dataset creation.
Create cleaner datasets for local intent-classification experiments.
Train local classifiers and keep model history organized.
Train a brand-new model or add a new training run to an existing model family. IntentOps Studio keeps dataset, metrics, model artifacts, and package exports connected.
Train ML.NET text classification models locally.
Track model families and training runs.
Export and import IntentOps model packages.
Prevent duplicate model and dataset package imports.
The full local intent-operations loop.
IntentOps Studio connects business context, conversation data, reviewed samples, datasets, models, evaluations, recommendations, and reports inside one desktop workspace.
Business Profile
Define business information, products, glossary terms, expected intents, and routing rules.
Conversation Import
Import structured JSON conversation exports from chatbot or support workflows.
Sample Review
Approve, correct, mark unknown, or discard messages before dataset creation.
Dataset Builder
Create intent datasets and export JSON, JSONL, CSV, or portable dataset packages.
Training
Train local ML.NET intent classifiers using reviewed datasets.
Evaluation
Inspect accuracy, Macro F1, Micro F1, LogLoss, confusion matrix, and per-intent metrics.
Recommendations
Find weak intents, dataset gaps, confusing classes, and next improvement actions.
Reports
Generate readable HTML reports for import analysis, datasets, model evaluation, and recommendations.
Export datasets, models, recommendations, and reports.
IntentOps Studio creates practical artifacts that can be archived, reviewed, shared, or reused in another project.
Business Profile JSON
Business context, products, glossary, intents, and routing rules.
Conversation Export JSON
Structured imported conversations with messages, channels, timestamps, and outcomes.
Dataset JSON / JSONL / CSV
Reviewed intent samples in structured, training-friendly, and spreadsheet-friendly formats.
Dataset Package
Portable IntentOps dataset archive with duplicate import protection.
Model Package
Portable trained model archive with manifest, metrics, and duplicate import protection.
Recommendations / Gaps JSON
Machine-readable improvement actions and coverage gaps for the selected model.
HTML Reports
Readable reports for import analysis, dataset summary, model evaluation, and recommendations.
Requirements and distribution.
The public release is packaged for Windows 10/11 x64 as a self-contained installer.
Operating system
Windows 10 or Windows 11 x64.
Runtime
.NET 8 self-contained build. No separate .NET installation is required for the release package.
Architecture
Windows x64.
Privacy
Imported conversations, datasets, model training, evaluations, recommendations, and reports run locally.
Source code
Available as an open-source project on GitHub under the MIT License.
Frequently asked questions
Questions before downloading?
IntentOps Studio is built for local intent-operations workflows, dataset preparation, model evaluation, recommendations, and business-readable reports.
IntentOps Studio is a Windows desktop app for importing customer conversations,
reviewing intent samples, building datasets, training local classifiers, evaluating model quality,
generating recommendations, and exporting reports.
No. The main workflow runs locally. Conversation import, dataset creation, ML.NET training,
evaluation, recommendations, and report generation happen on the user's machine.
IntentOps Studio imports structured conversation JSON files containing source metadata,
business information, conversations, messages, channels, timestamps, and outcomes.
Yes. Projects can be created with encrypted local project databases. Keep the project password safe,
because IntentOps Studio cannot recover lost passwords.
Small datasets are useful for demos and workflow validation, but production classifiers need more reviewed
examples per intent. IntentOps Studio helps identify weak intents and missing coverage.
The source code is available on GitHub at
github.com/micilini/IntentOps-Studio.
Ready to build a local intent-operations workflow?
Download IntentOps Studio for Windows 10/11 x64, create a local project, import conversations, review samples, build a dataset, train a model, and generate reports.