New

🎉 New product!

We’re excited to introduce Jobcurator, an open‑source Machine Learning library designed to clean, normalize, structure, compress, and sample large job datasets — before they reach your AI, ranking, or search systems.

Jobcurator acts as a data curation layer, sitting upstream of your pipelines to ensure job data quality from the very first step.

😍 Why it’s a big deal for HrFlow.ai users?

At HrFlow.ai, our mission is to solve unemployment, one API at a time — and high‑quality job data is a critical part of the equation.

If you work as a Data Scientist, Data Engineer, or Product Manager on a job board, aggregator, or programmatic distribution platform, you already know the problem:

• Job feeds are messy, highly redundant, poorly structured, and hard to scale efficiently.
• Raw job data is rarely ready for intelligent processing.

Jobcurator was built to fix this.

It transforms noisy job streams into high‑quality, deduplicated, and diverse datasets, ready to power:

• AI ranking & matching
• Job recommendation systems
• Search engines
• Analytics pipelines

Results: Better data in → better AI out.

✨ What’s included in this release?

Jobcurator provides a powerful yet lightweight toolkit to process job feeds at scale.

Core capabilities

• Clean & organize job data — remove duplicates and normalize information
• AI‑ready outputs — optimized for ranking, matching, and search systems
• Smart compression — keep the best jobs while preserving diversity
• Robust similarity detection — multi‑probe LSH with geo‑aware clustering
• Incremental processing — handle new batches over time

Flexible architecture

• Multiple backends available:
  • SimHash
  • MinHash
  • FAISS
• Optional outlier detection
• Supports local or SQL‑based storage for incremental processing

Built for scale

• Handles millions of jobs
• Memory‑friendly — no heavy infrastructure required
• No GPUs, no embeddings needed
• Tiny footprint — ~255 KB installed
• Optimized for large‑scale processing with minimal overhead

🔧 How does it work?

Jobcurator can be integrated easily into existing data pipelines.

At a high level:

1. Ingest raw job data (job feeds, aggregators, crawlers, APIs)
2. Choose a similarity backend (SimHash, MinHash, FAISS, etc.)
3. Run deduplication and smart compression
4. Output a clean, structured, and diverse job dataset
5. Send the curated data to your AI, search, or ranking systems

Jobcurator works out of the box, with sensible defaults — and can be progressively customized as your needs grow.

📎 Code examples and advanced configurations are available in the GitHub documentation.

💡Useful Links

• Jobcurator Full documentation on GitHub
• Sign up to HrFlow.ai
• API Authentification
• Create and configure a Board

🛣️ Our roadmap is public!

Looking for a specific feature or improvement?

Submit a request or upvote an existing one here:

👉 https://roadmap.hrflow.ai/

Update

🎉 New upgrade!

We’ve released a major upgrade to our Jobs Grading algorithms, delivering both higher accuracy and significantly faster processing when grading jobs for a profile.

This update improves two of our most advanced graders:

• grader-hrflow-jobs-cosmic
• grader-hrflow-jobs-titan

These improvements make job grading more reliable and more scalable for real‑world production use cases.

😍 Why it’s a big deal for HrFlow.ai users?

Grading jobs for a profile is a core building block for:

• Job recommendation engines
• Career guidance platforms
• Internal mobility tools
• Talent marketplaces

In these scenarios, two constraints are critical:

• The grading must be highly accurate
• It must also be fast enough to scale across large job catalogs

Until now, teams often had to compromise between precision and performance.

This upgrade removes that trade‑off.

📈 Accuracy improved

Both graders now reach 98.7% accuracy, measured using AUC (Area Under the Curve) on our internal benchmark datasets.

This leads to:

• More relevant job recommendations
• Better alignment between profiles and job requirements
• Higher trust in grading results

⚡ Speed massively improved

Job grading performance has been significantly accelerated:

• grader-hrflow-jobs-cosmic

  → 4.5× faster → 30 jobs scored/second

• grader-hrflow-jobs-titan

  → 7.5× faster → 50 jobs scored/second

These gains unlock smoother experiences when grading thousands or millions of jobs, even in near‑real‑time workflows.

Results : More accurate grading + faster execution = better job recommendations at scale.

✨ What’s included in this release?

This upgrade brings:

Improved grading quality

• 98.7% AUC accuracy on both Cosmic and Titan job graders
• Better discrimination between relevant and non‑relevant jobs

Performance gains

• Significant speed improvements across both graders
• Higher throughput for large‑scale job grading pipelines

Zero‑friction adoption

• No fine‑tuning required
• No configuration changes
• Fully backward‑compatible with existing integrations

🔧 How does it work?

No action is required to benefit from this upgrade.

To grade jobs for a profile:

1. Log in to the HrFlow.ai Portal
2. Activate the Job Grading API
3. Index jobs in a board and a target profile
4. Call the Job Grading endpoint using:
   • grader-hrflow-jobs-cosmic or
   • grader-hrflow-jobs-titan

All grading requests now automatically benefit from the improved accuracy and speed.

💡 Useful Links

• Job Grading API Endpoint
• API Endpoints Overview

🛣️ Our roadmap is public!

Looking for a specific feature or improvement?

Submit a request or upvote an existing one here:

👉 https://roadmap.hrflow.ai/

New

🎉 New Widget!

We’re thrilled to launch the AI Hiring Agent, a new AI‑powered widget designed to simplify and enhance recruitment workflows by automating candidate discovery, pre‑selection, and ranking from your existing talent sources.

Built as part of the HrFlow.ai AI Widgets ecosystem — alongside Recruiter Copilot and Talent Copilot — the Hiring Agent brings advanced AI recruitment intelligence directly into your ATS, CRM, or talent pool.

😍 Why it’s a big deal for HrFlow.ai users?

Recruiters and talent acquisition teams often spend significant time on early‑stage screening and candidate short‑listing, especially when working with large talent pools or internal databases.

The Hiring Agent solves this by:

• Automating pre‑selection

  It defines your candidate pool using criteria derived from a job offer or prompt text.

• Ranking with real‑time precision

  Profiles are examined in depth and ranked by relevance and likelihood of success giving recruiters a prioritized shortlist without manual sifting.

The result: fewer manual steps, faster identification of high‑potential candidates, and a more consistent, data‑driven hiring pipeline.

✨ What’s included in this release?

The AI Hiring Agent provides:

• Automated candidate search across indexed talent pools (ATS, CRM, or custom Source)
• AI‑driven ranking based on job and profile characteristics
• Real‑time ranking output, streamlining short‑listing
• A seamless widget experience in the HrFlow.ai Portal

This widget brings together multiple AI capabilities you already rely on — parsing, scoring, grading, and NLP — into a single, easy‑to‑deploy interface.

🔧 How does it work?

Getting started with the Hiring Agent is simple:

1. Log in to the HrFlow.ai Portal and go to UI Studio

2. Select Hiring Agent from the widget list

3. Click Create AI Widget and enter a name

4. Choose your profile Source (e.g., ATS or CRM)

5. Select the job Board where your job descriptions are stored

6. Choose appropriate algorithms (scoring + grading)

7. Set language and authentication levels

8. Launch your Hiring Agent — with a preview or generated link ready to use Once configured, the Hiring Agent will process your talent pool, search based on job criteria, and deliver a ranked list of matching profiles, ready for evaluation and outreach.

💡 Useful Links

• AI Hiring Agent Guide
• AI Hiring Agent Youtube Demo Video
• Sign up to HrFlow.ai

🛣️ Our roadmap is public!

Looking for a specific feature or improvement?

Submit a new request or upvote an existing one on our public roadmap:

👉 https://roadmap.hrflow.ai/

Update

🎉 New Upgrade!

We’ve released an upgrade to our Upskilling API at HrFlow.ai, improving how explanations are generated for profiles & jobs scores.

This update strengthens our commitment to explainable AI (XAI), making ranking results easier to understand and act upon.

Upskilling API Response

😍 Why it’s a big deal for HrFlow.ai users?

Ranking candidates and jobs is essential for recruiters, candidates, and employees — but scores alone are not enough.

At HrFlow.ai, we believe AI should never be a black box.

That’s why we built the Upskilling APIs: to clearly explain why a given score or recommendation was produced, in human‑readable language.

With this upgrade, explanations are now:

1. Clearer and More Focused

   We removed Opportunities and Threats, keeping only:

   • Strengths
   • Weaknesses

   This simplifies interpretation and reduces cognitive load.

2. More Structured and Machine‑Readable

   Each Strength or Weakness is now returned as an object with:

   • name
   • description

   instead of unstructured raw text.

3. More Context‑Aware

   • Profile explanations for a Job are now closer to the job requirements
   • Job explanations for a Profile are now closer to the profile’s actual skills and experience

The result: more relevant explanations, better transparency, and higher trust in AI‑driven recommendations.

✨ What’s included in this release?

This upgrade delivers:

• A simplified and clearer explanation structure
• Structured Strengths & Weaknesses objects
• More relevant explanations aligned with job or profile requirements
• Improved consistency across profile‑to‑job and job‑to‑profile explanations

🔧 How does it work?

You can start using the improved Upskilling APIs with no additional configuration.

To generate clearer explanations for scores and recommendations:

For Job explanations (candidate‑side)

1. Log in to the HrFlow.ai Portal
2. Activate the Job Upskilling API
3. Index a profile and a job
4. Call the API to explain why a job is recommended for a given profile

The API returns structured Strengths and Weaknesses that explain the job recommendation in clear, human‑readable terms.

For Profile explanations (recruiter‑side)

1. Log in to the HrFlow.ai Portal
2. Activate the Profile Upskilling API
3. Index a job and a profile
4. Call the API to explain why a profile is recommended for a given job

The API highlights strengths and gaps relative to the job requirements.

No changes are required for existing integrations — all users automatically benefit from clearer, more structured explanations.

Example of Upskilling API response:

{
    "code": 200,
    "message": "SWOT prediction finished in 19.86 seconds.",
    "data": {
        "strengths": [],
        "weaknesses": [
            {
                "description": "Aucune expérience ou compétence en gestion de plateforme numérique, UX, Compass ou Waka Well n'est mentionnée.",
                "name": "Compétences en gestion de plateforme (Waka Well, Waka Well 2.0, UX, Compass)"
            },
        ],
    },
}

💡 Useful Links

• Job Upskilling API Endpoint
• Profile Upskiling API Endpoint
• Sign up to HrFlow.ai
• API Authentification
• HrFlow.ai Python SDK

🛣️ Our roadmap is public!

Looking for a specific feature or improvement?

Submit a new request or upvote an existing one here:

👉 https://roadmap.hrflow.ai/

Update

🎉 New update!

We’ve released an update to the Asking API at HrFlow.ai, delivering higher accuracy and improved handling of accentuated characters across all supported languages.

This upgrade further strengthens our LLM‑powered API designed to interact intelligently with profiles and jobs.

😍 Why it’s a big deal for HrFlow.ai users?

The Asking API is an LLM trained on HR‑specific data, built to help users interact naturally with CVs and job descriptions.

For example, recruiters can:

• Ask a profile (CV):

  “I’m meeting this candidate for a junior data scientist position. Give me 5 questions to assess their skills.”

• Ask a job description:

  “Translate this job description into Japanese.”

  “Correct typos and make it more attractive for candidates.”

Thanks to users feedback, this update significantly improves the quality of responses:

1. Higher Answer Accuracy

   Accuracy has been improved by 30% (from 68.25% → 77.78%) on our internal benchmark.

2. Better Handling of Accentuated Characters

   Accentuated characters are now correctly rendered in API outputs across all 43 supported languages.

The result: more reliable answers, cleaner outputs, and a better experience for recruiters and candidates alike.

✨ What’s included in this release?

This update delivers:

• Improved LLM accuracy for Profile & Job Asking
• Correct rendering of accentuated characters
• Full support across 43 languages
• No breaking changes to existing integrations

🔧 How does it work?

Job Asking

To ask questions about a job follow the guide here or below:

1. Log in to the HrFlow.ai Portal
2. Activate the Job Asking API
3. Create and configure a Board
4. Index a job in a board
5. Ask your question via the API

Profile Asking

To ask questions about a profile follow the guide here or below:

1. Log in to the HrFlow.ai Portal
2. Activate the Profile Asking API
3. Create and configure a Source
4. Index a profile in a source
5. Ask your question via the API!

No additional configuration is required — existing users benefit automatically from the improvements.

Ask questions to a Profile indexed in a Source

Ask questions to a Job indexed in a Board

💡 Useful Links

• Sign up to HrFlow.ai
• API Authentification
• HrFlow.ai Python SDK
• Create and configure a Source
• Our Developers’ Profile Asking Guide
• Create and configure a Board
• Our Developers’ Job Asking Guide

🛣️ Our roadmap is public!

Looking for a specific feature or improvement?

Submit a request or upvote an existing one on our public roadmap:

👉 https://roadmap.hrflow.ai/

New

🎉 New feature!

We’re excited to introduce a brand‑new Job Scoring algorithm from HrFlow.ai, designed to rank jobs for a profile at massive scale — with +90% accuracy out of the box, and no fine‑tuning required.

This is a major breakthrough in job recommendation industry, making high‑accuracy scoring instantly accessible.

😍 Why it’s a big deal for HrFlow.ai users?

Job ranking is a core challenge for career platforms, internal mobility tools, and job boards.

Until now, achieving high‑quality job recommendations required:

• Using a fast scoring model with ~70% default accuracy
• Investing time and expertise into fine‑tuning
• Relying on large volumes of high‑quality training data

This new algorithm changes the game.

With this update, users can now:

1. Access +90% Accuracy Instantly

   Get high‑quality job rankings without any retraining or configuration.

2. Rank Millions of Jobs in Seconds

   Designed for One‑to‑Many use cases, this algorithm efficiently ranks massive job catalogs against a single profile.

3. Deliver Smarter Job Recommendations

   Help candidates or employees quickly discover the most relevant opportunities on career sites and internal platforms.

4. Simplify Job Recommendation Pipelines

   No fine‑tuning, no complex setup — just reliable, high‑performance scoring out of the box.

The result: better job suggestions, faster implementation, and higher user satisfaction.

✨ What’s included in this release?

This update introduces:

• A new Job Scoring algorithm optimized for profile‑to‑jobs ranking
• +90% accuracy by default, with no fine‑tuning required
• Support for 43 languages
• Ability to score millions of jobs in seconds
• Seamless availability through the Algorithms Marketplace

🔧 How does it work?

To start using the new Job Scoring algorithm:

1. Log in to the HrFlow.ai Portal
2. Navigate to AI Studio
3. Select Algorithms Marketplace
4. Go to Scoring Algorithms (One-to-Many)
5. Click on AI Job board
6. Use the algorithm to rank jobs for a given profile via the Job Scoring API

AI Studio > Algorithm Marketplace > Scoring Algorithms (One-to-Many) > AI Job board

💡 Useful Links

• Sign up to HrFlow.ai
• API Authentification
• Our Developers Profile Scoring Guide
• Our Developers Job Scoring Guide

🛣️ Our roadmap is public!

Looking for a specific feature or improvement?

Submit a request or upvote an existing one on our public roadmap:

👉 https://roadmap.hrflow.ai/

New

🎉 New feature!

We’re thrilled to introduce a no‑code Recruiter Copilot generator in the HrFlow.ai Portal, enabling users to create their own AI‑powered talent pool widget in just a few steps.

This new experience gives teams full control to build, customize, and deploy Recruiter Copilots — without writing a single line of code.

😍 Why it’s a big deal for HrFlow.ai users?

Recruiter Copilots are AI‑powered widgets designed to help recruiters quickly identify the best candidates within their talent pool.

With this update, users can now:

1. Build Copilots Without Code

   Create a fully functional Recruiter Copilot through a guided, no‑code interface.

2. Customize the Copilot to Each Use Case

   Adapt the Copilot to different recruiters, teams, or workflows in just a few clicks.

3. Control Both AI Logic and UI

   Configure how candidates are selected, ranked, filtered, and displayed — all from one place.

The result: faster setup, more tailored experiences, and higher adoption by recruiters.

✨ What’s included in this release?

The Recruiter Copilot generator introduces full customization across five key areas:

1. Artificial Intelligence Settings: Configure the boards of jobs, sources of profiles, and the scoring algorithm.

2. General Settings: Set the language, authentication type, widget validity duration, and more.

3. Grid Design: Customize the tabs displayed in the widget, widget colors, and the custom tags shown in the profile list table.

4. Filters Design: Define your profile search parameters, such as location, seniority, keywords, and tags.

5. Profile Details Design: Customize the layout and content of the profile details page.

🔧 How does it work?

To create your own Recruiter Copilot:

• Log in with your company credentials or sign up to create an account.
• Go to UI Studio
• Select Widgets Marketplace
• Create a new Recruiter Copilot
• Configure the AI logic, design and filters
• Deploy your widget

UI Studio > Widgets Marketplace > Recruiter Copilot

Recruiter Copilot > Create a widget

💡 Useful Links

• Sign up to HrFlow.ai
• HrFlow.ai Python SDK
• API Authentification
• Recruiter Copilot Guide
• Recruiter Copilot Live Demo

🛣️ Our roadmap is public!

Looking for a specific feature or improvement?

Submit a request or upvote an existing one on our public roadmap:

👉 https://roadmap.hrflow.ai/

New

🎉 New feature!

We’ve added two new capabilities to the UI Studio in the HrFlow.ai Portal:

the Widgets Marketplace and the Widgets List.

Together, these features make it easier to create, discover, and manage widgets tailored to different users, data sets, and use cases.

We’ve added two new features to the UI Studio, available in the HrFlow.ai Portal: the Widgets Marketplace and the Widgets list.

😍 Why it’s a big deal for HrFlow.ai users?

Widgets are off‑the‑shelf automations designed to cover common recruiting and talent workflows.

They combine:

• Data (Sources & Boards)
• AI modules (Parsing, Scoring, Matching, etc.)
• User Interface (Hiring Agent, Recruiter Copilot, Talent Copilot, etc.)

Depending on the use case, the audience, or the data, users often need different widgets.

This update gives them the flexibility to do exactly that.

With these new interfaces, users can now:

1. Discover & Create Widgets Faster

   Browse available widgets and create new ones from a dedicated marketplace.

2. Manage All Widgets in One Place

   Access, edit, and organize all your widgets from a single list.

3. Adapt Widgets to Each Use Case

   Build different widgets for recruiters, hiring managers, candidates, or internal tools — without friction.

The result: faster setup, clearer organization, and more scalable UI automation.

✨ What’s included in this release?

This update introduces two complementary interfaces:

Widgets Marketplace

• Browse available widgets
• Create new widgets from scratch

Widgets List

• View all your created widgets
• Access and update existing widgets

Both are accessible directly from UI Studio.

🔧 How does it work?

To access the Portal and use our new features :

1. Log in with your company credentials or sign up to create an account
2. In the section UI Studio either click on :
   1. Widgets Marketplace : to create a new widget
   2. My widgets : to see or update an existing one

UI Studio > Widgets Marketplace

UI Studio > My Widgets

💡 Useful Links

• Sign up to HrFlow.ai
• HrFlow.ai Python SDK
• API Authentification
• Recruiter Copilot Guide
• Recruiter Copilot Live Demo
• Our Talent Copilot Guide
• Try our Talent Copilot

🛣️ Our roadmap is public!

Looking for a specific feature or improvement?

Submit a request or upvote an existing one here:

👉 https://roadmap.hrflow.ai/

Update

🎉 New upgrade!

We’ve released a major upgrade to our Tagging API with the launch of ROME 4.0 V2 taggers, the next generation of occupational classification models from HrFlow.ai.

This new version significantly improves performance, accuracy, and language coverage, making ROME‑based tagging more scalable and reliable than ever.

😍 Why it’s a big deal for HrFlow.ai users?

Following the successful release of ROME 4.0 last October, this V2 upgrade takes occupational tagging one step further with major improvements:

• ⚡ 21% faster than the previous ROME 4.0 version
• 🌍 43 languages supported (vs. 2 previously)
• 🎯 More accurate and relevant tag predictions
• 🇫🇷 Fully aligned with the “France Travail” referential

The result: faster classification, broader international coverage, and higher‑quality occupational tagging for jobs and profiles.

✨ What’s included in this release?

This upgrade introduces four new ROME 4.0 V2 taggers, covering the full ROME hierarchy:

• tagger-rome4-jobtitle2
• tagger-rome4-category2
• tagger-rome4-subfamily2
• tagger-rome4-family2

These new taggers are designed to replace or complement the previous ROME 4.0 versions, offering improved performance out of the box.

🔧 How does it work?

To start using the ROME 4.0 V2 taggers:

1. Activate the Text Tagging API
2. Send your job descriptions or profile texts to the API
3. Select one or more ROME 4.0 V2 taggers in your request:
   • tagger-rome4-jobtitle2
   • tagger-rome4-category2
   • tagger-rome4-subfamily2
   • tagger-rome4-family2
4. Retrieve standardized ROME tags with improved speed and accuracy

No additional configuration is required — existing tagging pipelines can adopt V2 seamlessly.

💡 Useful Links

• Live demo of Text Tagging
• Text Tagging API Endpoint
• Text Tagging API Guide
• API Endpoints Overview

🛣️ Our roadmap is public!

Looking for a specific feature or improvement?

Submit a request or upvote an existing one here:

👉 https://roadmap.hrflow.ai/

New

🎉 New feature!

We’re excited to introduce the Scoring Algorithms Marketplace in the HrFlow.ai Portal — a centralized interface to discover, test, and deploy scoring algorithms capable of ranking millions of profiles or jobs in seconds.

This new marketplace gives users access to a wide range of pre‑trained scoring algorithms, organized by job type, profile type, sector, and industry.

😍 Why it’s a big deal for HrFlow.ai users?

Scoring at scale is critical for both recruiters and candidates. At HrFlow.ai, we’ve built two foundation scoring algorithms:

• One to rank profiles against a job
• One to rank jobs against a profile

With an accuracy of up to 90%, these pre‑trained algorithms are designed to deliver high‑quality rankings out of the box, and can be fine‑tuned to match specific hiring or application strategies.

With the Scoring Algorithms Marketplace, users can now:

1. Access Ready‑to‑Use Scoring Algorithms

   Explore a catalog of high‑performance scoring models built for real‑world recruiting and job discovery use cases.

2. Prioritize at Scale

   Help recruiters focus on the most relevant candidates, or help candidates prioritize the best job opportunities.

3. Adapt Scoring to Your Context

   Fine‑tune scoring algorithms to reflect your data, market, or business logic.

The result: faster prioritization, better decision‑making, and scalable relevance — on both sides of the market.

✨ What’s included in this release?

The Scoring Algorithms Marketplace provides:

• A centralized catalog of Scoring algorithms
• Filtering by use case
• Real‑time testing of scoring models
• In‑context documentation to learn how to use the APIs
• Options to fine‑tune algorithms
• Seamless integration into your workflows via the Scoring APIs

All accessible directly from AI Studio.

🔧 How does it work?

To explore and use the Scoring Algorithms Marketplace:

1. Log in to the HrFlow.ai Portal
2. Navigate to AI Studio
3. Select “Algorithms Marketplace”
4. Choose Scoring Algorithms ( One-to-Many )
5. Select a model based on your needs

AI Studio > Algorithms Marketplace > Scoring Algorithms ( One-to-Many )

Algorithms Marketplace > Scoring Algorithms ( One-to-Many ) > AI Job board

Scoring Algorithms ( One-to-Many ) > AI Job board > Create Algorithm

💡 Useful Links

• Sign up to HrFlow.ai
• Our Developers’ Profile Scoring Guide
• Our Developers’ Job Scoring Guide
• HrFlow.ai Python SDK
• API Authentification

🛣️ Our roadmap is public!

Looking for a specific feature or improvement?

Submit a request or upvote an existing one here:

👉 https://roadmap.hrflow.ai/