ZivaBasa (ChiedzaAI)

The description states that ZivaBasa is a multi-task AI platform that forecasts employment impact, workforce skills gaps, and productivity shifts using a shared representation trunk feeding three task heads (employment, productivity, skills). It uses TensorFlow/Keras for modeling, SHAP for explainability, FastAPI for serving, and a React-based dashboard for human-facing interaction.

It is described as a first module in a broader AI platform called ChiedzaAI, with future modules including Cognify (training and reskilling) and NeuroWorks (orchestration).

The authors also state that the current version uses public Kaggle datasets as proxies, not real Zimbabwean banking data. The system is built to be explainable — showing why a forecast was made, not just what it is.

Inference The product is a research-grade prototype with a defined architecture and technical stack, but no evidence of deployment or production use beyond the hackathon phase.

The description states that ZivaBasa is built to address a gap in workforce forecasting tools, especially in African economies where institutions lack reliable datasets and forecasting capabilities. It positions itself as an explainable AI platform that can be understood, challenged, and acted on by regulators, executives, or employees.

It claims to offer:

• Joint forecasting of employment, skills, and productivity
• SHAP-based explainability
• Multi-task learning architecture

The authors also state that the platform is designed to avoid being a black box, emphasizing that it must be interpretable for stakeholders in decision-making roles.

Inference The positioning is rooted in a narrative of social impact and explainable AI, but there is no evidence of market traction or adoption by real users. The claims are self-reported and unverified.

The description states that ZivaBasa is focused on the banking sector, with an initial use case in Zimbabwe. It is designed for:

• Bank executives
• Regulators
• HR decision-makers

It also mentions that the platform aims to help institutions prepare for AI-driven workforce changes, especially in a context where there’s a risk of African economies becoming consumers of foreign AI platforms.

Inference The ICP appears to be public and private sector institutions in emerging markets with workforce planning needs. However, no evidence of actual customers or use cases is provided.

There is no evidence in the description of a business model or pricing strategy. The authors do not mention:

• Revenue streams
• Customer acquisition plans
• Pricing tiers
• Monetization approach

The project is described as a hackathon MVP, and no commercialization plan or monetization strategy is evident.

Inference No business model or pricing evidence is presented — this remains unaddressed in the self-reported description.

The technical stack includes:

• Frameworks: TensorFlow/Keras, FastAPI, SHAP, scikit-learn, XGBoost
• Frontend: React, Tailwind
• Backend: Python, Node.js, Uvicorn, Vite
• Data tools: Pandas, NumPy, MLflow, Jupyter, Parquet, PyArrow, Pydantic

The architecture is described as:

• A multi-task deep learning network with a shared trunk and three output heads (employment, productivity, skills)
• SHAP-based explainability layer
• FastAPI serving endpoints
• Prototype dashboard with batch upload, KPI lookups, and conversational interface

The authors also note that they benchmarked against classical baselines to validate the multi-task model’s complexity.

Inference There is a clear technical architecture and implementation plan. However, no evidence of production deployment or scalability beyond the MVP phase.

The description states:

• The project was built in 14 days as a hackathon sprint
• It uses proxy data from Kaggle, not real Zimbabwean banking data
• It includes a prototype dashboard and API endpoints
• It is transitioning to real bank partnerships and federated learning

However, there is **no evidence of:

• Real customers or users
• Revenue or monetization
• Product adoption or usage metrics
• Deployment in production environments

The authors explicitly state that the current phase validates pipeline mechanics, not findings.

Inference This is a pre-product, pre-traction prototype. No maturity or traction signals are evident.

There is no mention of competitors in the description. The authors do not reference:

• Existing AI workforce forecasting platforms
• Competing tools in the explainable AI space
• Similar solutions in the banking or HR analytics domain

The project does not appear to be positioned against any known market players.

Inference No competitive context is provided — this leaves open the question of how ZivaBasa differentiates from existing tools, if any.

• Proxy data dependency: The MVP uses public datasets, not real-world data. This limits the validity of predictions and raises questions about scalability.
• No commercialization or monetization strategy: No evidence of a business model or customer base.
• Single-founder team: The project is built by one person (Mark Chindudzi), which may limit execution capacity.
• Technical bugs noted: Silent schema mismatches and data leakage issues were found during development, suggesting potential instability in the system.
• Unproven transition to real-world use: The roadmap includes moving to real partnerships and federated learning, but no evidence of progress or success in those areas.

Inference The project is at a very early stage with significant risks around scalability, commercial viability, and technical robustness.

1. What are the actual data sources you plan to use for real-world deployment? Are they private or public?
2. Have you begun any pilot partnerships with banks or institutions in Zimbabwe or other African markets?
3. How do you plan to validate that your model's predictions hold up outside of proxy datasets?
4. What is your roadmap for transitioning from SHAP-based explanations to causal-consistent explainability?
5. Do you have a plan for deploying the system at scale, including containerization and CI/CD?
6. What are the key assumptions in your multi-task architecture that could break down in real-world use cases?
7. How do you intend to monetize this platform? Are there any revenue streams already in place?