Predictive
What happens if nothing is done? Strata projects risk before it becomes a funding loss.
AI infrastructure for municipal intelligence
Infrastructure for auditable municipal AI.
Hub Esfera builds Strata Engine, an AI-native municipal intelligence platform for Brazilian municipalities. Strata turns public datasets, fiscal rules, legal sources, and municipal indicators into auditable recommendations that help public managers identify missed or at-risk funding before regulatory deadlines close.
Brazil has 5,570 municipalities. Most lose public funding not because the money does not exist, but because risks, eligibility gaps, regulatory deadlines, and execution paths are detected too late.
Production focus
Legal retrieval Conversational intelligence Risk scoring Evaluation pipelines Traceable recommendations
§ 01 · What we are building
Public-sector intelligence for municipal funding.
Strata Engine is a public-sector intelligence layer for municipal funding, fiscal risk, and regulatory decision-making. Each recommendation includes its underlying data, legal basis, and decision path, making the output auditable for public managers, internal control teams, courts of accounts, and institutional partners.
Preventive
What needs to be organized before the next regulatory window closes?
Prescriptive
What plan should the municipality execute now, with a traceable legal and operational path?
§ 02 · Why this is an AI workload
Every city analysis becomes structured AI work.
As Strata scales from one city to thousands, the system must ingest public datasets, retrieve legal and fiscal context, evaluate eligibility rules, score funding risks, detect deadline-sensitive opportunities, and generate explainable recommendations with source traceability.
- ✓Legal retrieval over laws, decrees, ordinances, transfer rules, and official sources
- ✓AI-assisted reasoning over fiscal, regulatory, and municipal context
- ✓Conversational intelligence for municipal decision-making, enabling follow-up questions and decision support
- ✓Risk scoring for missed funding, eligibility gaps, and deadline-sensitive opportunities
- ✓Document understanding for public-sector rules and funding programs
- ✓Batch analysis across Brazilian municipalities
- ✓Tenant-isolated context and memory for each city, user, thread, and workflow
- ✓Evaluation pipelines for accuracy, traceability, hallucination control, and legal consistency
- ✓Low-latency inference for interactive municipal intelligence workflows
§ 03 · Current stage
Bootstrapped, working, and moving toward repeatable coverage.
Hub Esfera is bootstrapped, with no external capital raised. We have a working Strata PoC, 11 public data sources integrated, a public-sector launch motion in Porto Alegre, and a documented 3P method: Predictive, Preventive, and Prescriptive intelligence.
Working PoC
Strata already demonstrates municipal analysis over public-sector data and funding rules.
Launch motion
The current go-to-market motion starts in Porto Alegre with public-sector institutional relationships.
Scaling goal
Move from single-city analysis to repeatable multi-municipality coverage without losing auditability or trust.
§ 04 · Why infrastructure partners matter
Not generic cloud capacity. Production AI infrastructure.
We are building a production AI workload for public-sector intelligence in one of the world's largest municipal markets. Infrastructure support would help us accelerate the workloads that determine whether Strata can scale with rigor.
Inference and retrieval
LLM inference, legal and regulatory retrieval, open-source model evaluation, and document understanding.
Municipal scale
Multi-city batch processing, risk scoring, eligibility analysis, and deadline-sensitive opportunity detection.
Government trust
Secure, auditable AI workflows for government use, with tenant-isolated context and a path toward smart-city simulation workloads.
§ 05 · Smart Cities R&D
Real-world pain. Synthetic data. Public value.
Hub Esfera's R&D roadmap explores how synthetic, non-identifying citizen journeys can help municipalities evaluate the impact of public decisions before execution. Strata identifies the funding opportunity, regulatory risk, or recommended intervention. The simulation layer asks what changes for citizens when that decision becomes a service.
Decision trigger
Strata identifies a funding opportunity, regulatory risk, or recommended public-service intervention.
Representative journey
A future simulation layer evaluates how the decision may affect a citizen archetype such as Dona Maria.
Public value
The goal is to help cities make better decisions before those decisions reach citizens.
§ 06 · Why Hub Esfera
Domain knowledge meets AI-native execution.
We combine public-sector domain knowledge, municipal data pipelines, legal and fiscal traceability, AI-assisted reasoning, tenant-isolated institutional context, a working GovTech PoC, and a long-term R&D path toward citizen-centered smart cities. We are building a repeatable product layer, not a one-off advisory workflow.
- ✓Public-sector domain knowledge
- ✓Municipal data pipelines
- ✓Legal and fiscal traceability
- ✓AI-assisted reasoning
- ✓Tenant-isolated institutional context
- ✓A working GovTech PoC
Partner contact
Discuss AI infrastructure partnership.
For inference, cloud, model, evaluation, or smart-city infrastructure providers. Share the support, credits, program, or integration path that may fit Strata.
LLM inference Legal retrieval Model evaluation Smart cities R&D