First Public Release: 2026-04-08
Last Updated: 2026-04-09
Automated Agile Task Force Optimization using Heterogeneous Graphs
AgileTaskHeron is an intelligent graph engine that automatically prioritizes tasks and creates optimal Task-Skill-Person bundles for Agile teams.
By modeling Tasks, Required Skills, and Human Resources (including past experience) in a single heterogeneous graph, it delivers:
- Task priority ranking using PageRank
- Natural team bundles using Louvain community detection
- Ready-to-use Agile sprint recommendations
Built with GPU acceleration (RAPIDS cuGraph + cuDF) for high performance.
- Task-centric priority calculation (PageRank)
- Automatic discovery of meaningful Task-Skill-Person bundles (Louvain)
- Weight-based importance control (HR team’s subjective judgment supported)
- GPU-accelerated computation
- Easy-to-use Jupyter Notebook demo
While PageRank and Louvain community detection are well-known algorithms used in many graph analysis tools, AgileTaskHeron is fundamentally different in purpose and design.
Key differentiators:
-
Task-centric heterogeneous graph
Most tools analyze general networks (social, knowledge, or organizational graphs).
AgileTaskHeron is specifically designed around Tasks as the central node, connecting Required Skills and Human Resources in one unified graph. -
Task-first priority + natural bundle generation
It doesn’t just find communities or important nodes — it simultaneously calculates which tasks should be done first and automatically creates ready-to-use Task-Skill-Person bundles. -
No training required
Unlike deep learning or machine learning-based team recommendation systems, AgileTaskHeron works instantly using only your current data and edge weights. No historical dataset or retraining is needed. -
Full user control over edges and weights
You define every relationship and importance (weight) yourself. The algorithm respects your domain knowledge and business priorities completely. -
Built for real Agile / HR workflows
The output is not abstract graph metrics — it directly produces Agile sprint priority + recommended task forces that HR and Agile teams can use immediately.
In short, while the underlying algorithms are standard, the combination, focus on Tasks, and practical Agile application make AgileTaskHeron unique.
Other tools may use PageRank + Louvain for general network analysis, but none combine them in this specific way to solve dynamic Agile Task Force formation.
You already have an AI team.
You already have a UI development team.
You already have GPU resources (or can easily spin them up in the cloud).
Yet you're still paying ~$8,724 per year for Jira® Premium (50 users) — and sending your company's internal plans, strategic priorities, and sensitive data to a third-party cloud.
Jira® is excellent for basic issue tracking.
Jira® AI is nice for writing summaries and suggestions.
But neither of them can solve the real problem Agile teams face every sprint:
"Which tasks are truly the most important right now, and who should actually work together on them?"
AgileTaskHeron was built precisely for companies like yours.
- Runs entirely on your own infrastructure — your data never leaves your control
- No training, no historical data required — works immediately with your existing knowledge
- Uses graph intelligence (PageRank + Louvain) to automatically create optimal Task-Skill-Person bundles and task priorities
- Gives you something Jira® AI still can't: real, actionable Agile team formation
In short:
You don’t need another ticket system.
You need a smart optimization engine that actually makes your Agile delivery faster and smarter.
Unlike deep learning-based team recommendation systems that require extensive historical data, long training periods, and periodic retraining whenever the organization changes, AgileTaskHeron is completely training-free.
It is designed to work instantly using only your existing data:
- Leverages existing knowledge: Team leaders’ performance reviews, past project evaluations, and domain expertise can be directly converted into edge weights.
- Instant Task Force creation: As soon as a new task appears, the system can immediately generate optimized Task-Skill-Person bundles and priorities — no retraining needed.
- Truly Task-Oriented: It explicitly calculates which tasks are most important and in what order they should be executed, rather than simply matching people to skills.
- High flexibility: Users have full control to define and adjust edge weights based on business priorities, strategic importance, or real-world experience.
This makes AgileTaskHeron particularly powerful for HR teams and Agile organizations that want fast, practical, and explainable results without the complexity and maintenance overhead of deep learning models.
conda env create -f environment.yml
conda activate rapidspython agile_task_bundle.py- Prepare your edges DataFrame with src, dst, edge_type, and weight
- Call analyze_task_bundle(edges)
- Get Agile priority order + recommended bundles instantly
Weight Guidelines
- Range: 0.1 ~ 1.0 (0 is not allowed)
- 1.0 = critically important
- 0.1 = slightly related
The weight is subjective — HR or Agile teams should assign values based on real business importance, past performance, and strategic priority.
To the best of our knowledge, no existing patent covers the specific combination of PageRank-based task priority ranking and Louvain community detection applied to a Task-Skill-Person heterogeneous graph for automated Agile Task Force formation. Searched across USPTO, EPO, WIPO, KIPRIS, and CNIPA via Google Patents, Espacenet, KIPRIS, and Patentscope. (April 2026)
2000 persons, 50 tasks: Total Edges: 78,935
Took 1.9 sec on AWS g4dn.xlarge
On-demand price: ~$0.526/hour — a weekly sprint run costs less than $0.01.
See more from practical-example.ipynb
AGPL v3 — free for research and non-commercial use.
- Personal use
- Educational purposes (university courses, student projects, bootcamps, workshops, academic research)
- Non-commercial research and open-source projects
Commercial use requires a separate agreement.
- Internal company use
- SaaS products or any for-profit applications
- Government and public sector institutions (including national research institutes, public agencies, etc.)
- Requires a separate Commercial License
Special Note for SI Companies
We welcome System Integrators (SI companies) who are interested in building a more cost-effective Jira-like solution or launching their own SaaS product using AgileTaskHeron. Commercial licensing and partnership opportunities are available.
For commercial licensing, enterprise support, or partnership inquiries, please leave a message in Discussions.
For commercial / closed-source use:
| Users | Annual License |
|---|---|
| up to 50 | $2,000 / year |
| up to 100 | $3,500 / year |
| up to 200 | $4,500 / year |
| Enterprise (unlimited) | Custom quote |
| Jira® Premium | AgileTaskHeron | |
|---|---|---|
| Up to 50 users / year | ~$8,724 | $2,000 |
| Up to 100 users / year | ~$17,448 | $3,500 |
| Up to 200 users / year | ~$34,896 | $4,500 |
| Enterprise (unlimited) | Custom | Custom quote |
| Graph-based automatic team formation | Not included | ✅ |
| Automatic task priority ranking | Not included | ✅ |
| Data privacy | Atlassian cloud | Your own server |
| Infrastructure | Included | Your own (AWS, on-prem, etc.) |
Note: The Core Engine License is intended for teams that already have their own internal UI, DevOps, and integration capabilities.
The prices above apply to the Core Engine License only.
For teams that need a ready-to-deploy, self-hosted AgileTaskHeron system, including Docker deployment, web dashboard, enterprise connectors, or on-premise integration support, please contact us for the Professional Docker Edition or Enterprise Edition.
Jira® is a registered trademark of Atlassian Corporation.
Pricing verified via official Atlassian pricing page, April 2026.
Copyright © 2026 Klastrovanie Co., Ltd. All rights reserved.