01 · Data FoundationIn production
Connect, structure and contextualize enterprise data.
ERP systems, databases, operational data, existing models and algorithms in one traceable decision layer, inside your infrastructure.
Decision-Grade AI
Anticipate financial threats before you commit.
Grand Thera builds decision-grade AI that separates evidence from noise, quantifies what remains uncertain and exposes the financial impact before it reaches the P&L.
Tap a city to open its exposure report
Beyond Ontology · Working Paper v1.1 · DOI 10.5281/zenodo.22240288RedMind runs on TheraOS · 60+ AIT Agents in real timeDeployed inside the client's infrastructure · no external data accessL0–L9 layered inference architectureTroIA in closed testing with selected clientsGDPR and LGPD by default
Computational epistemology
Watch an inference form.From noise to a decision you can defend.
Ontology structures what is available to the machine. Computational epistemology structures what the machine is permitted to conclude from it. Scroll to run the path.
Why we are here
We exist because reports arriveafter the damage is done.
Information is abundant. Direction is not. Most systems report the past, so decisions come after exposure has already moved. We deploy proprietary decision-grade AI that delivers real-time direction and execution, reducing surprise exposure before it hits the P&L.
Fragmented data → one decision layer
Unified
TheraOS unifies ERP systems, databases, models and algorithms into a single source of truth, deployed inside your infrastructure.
Late reports → direction before impact
Anticipatory
Regime shifts and propagating risk are detected while there is still time to act, before they reach the P&L.
Overconfidence → evidence that survives
Calibrated
Computational epistemology decides which data deserves belief: every inference carries its probability and confidence.
Black boxes → decisions you can defend
Accountable
AIT Agents prescribe and execute with an owner, lineage and the evidence behind each recommendation.
Risk contagion
Financial threats rarely announce themselves.They propagate.
An FX move and a macro shock become production cost and supplier stress, then a credit, liquidity and margin problem. Scroll to watch it spread, then rewind and run it again with TheraOS in the loop.
Predictive capabilities
See it coming.Quantify it. Explain it. Act on it.
Reports describe the past. TheraOS detects the early signal, anticipates where the indicator is heading, quantifies the probability of breaching the board's limit and the exposure behind it, explains the drivers and turns them into a response with an owner. Choose a scenario and change its parameters.
Decision workbench
WatchAn early signal, before it shows in a report.
Delinquency in one sector is rising faster than its seasonal pattern.
- Detected
- Quantified
- Explained
- Responding
- Probability of breach
- —
- in 13 weeks
- 80% interval
- —
- index at +13 w
- Exposure at risk
- —
- of R$ 420 M
- Lead time
- —
- until breach is likely
Adverse-event likelihood · week +13
—
Each dot represents 5%Model confidence 81%
Probability of breach over time
50%Available at the Quantify stage
+1 w+13 w
Illustrative simulation: data, probabilities and amounts are generated from the parameters above to show how the stages work. They are not customer results or live production forecasts.
How TheraOS works
From fragmented datato a baseline the board can govern.
We deploy our decision-grade AI stack inside your infrastructure: TheraOS reconciles the data, the board sets the risk appetite, AIT Agents enforce it, and every decision is measured against a financial baseline.
TheraOS · Financial Risk Baseline
The baseline the board governs.In the product, today.
Critical indicators, exposure outside the baseline, baseline evolution and the AIT actions that bring it back, in one governed view. Explore each element of the real interface.
01 / 09
Financial Risk Baseline
Continuous governance of energy, contract and operating-cost exposure: the expected financial trajectory every decision is held against.
TheraOS product interface · demonstration environment.
TheraOS
In productionScientific AI inferencefor mission-critical decisions.
The operational core of our decision-grade AI stack: enterprise data, specialized models, risk analysis and decision workflows in one governed layer. Deployed inside your infrastructure, with no external data access.
Explore Data Foundation
02 · Command CenterIn production
Exposure, uncertainty and the decision in one view.
Signals by state, each with its recommendation, impact, exposure, confidence, owner and evidence freshness.
03 · AIT AgentsIn production
From signal to probability, exposure and action.
Purpose-deployed agents that recommend steps, owners and timing, and execute pre-authorized responses under your control.
04 · TroIAIn development
A coordinated response when the regime shifts.
Regime-shift detection and orchestration of AIT Agents across cash flow and key indicators. In closed testing.
Not ready for a full integration? Zero-Data Onboarding begins with the decision you need to make and the minimum evidence it requires.
Zero-Data OnboardingTheraOS Advisor
Ask it the way you would ask an analyst.It answers with evidence.
The natural-language interface to TheraOS. Ask about exposure, scenarios or an alert and get a probability, the financial impact, an owner, a recommended response and the sources behind it. Try a question below, or open a city on the map and send its report here.
Loading the TheraOS Advisor
In production
Decision-Grade AI in production.Epistemic rigor where capital is at stake.
Not a pilot. Our decision-grade AI stack runs in production for clients whose decisions carry real financial consequence, and every inference it delivers has to survive the evidence before anyone acts on it.
Trusted by leading organizations
In their words.Leaders who decide with it.
“What Grand Thera has built with Decision-Grade AI addresses a core executive challenge: acting with clarity under uncertainty.”
Guga StoccoTech Investor & CEO, RedMind“Grand Thera demonstrates rare technical depth and their technology handles complex decision scenarios with real precision.”
Tiago AlvesBanker, BTG Pactual Advisors“With Grand Thera's Decision-Grade AI, we accelerated new projects by turning data into usable decisions much faster.”
Leonardo GoldimFounder, Virtual Officer & IT2Group
Applications
Mission-critical,where error is expensive.
Loading applications
TheraOS
Scientific AI, engineeredto drive the best financial outcomes.
Deployed inside your infrastructure, for the moments where error is expensive and visible.
The Grand Thera Manifesto
Read the manifestoWhile most AI applications focus on consumer use, we are building AI for a world defined by uncertainty. Our technology is designed for critical decision-making, empowering enterprises that are redefining how markets function and shaping what comes next.
Evidence laboratory
A model can fit anything.Evidence has to survive.
Change the quality of the signal, how often it persists and how much three model families share. Watch the belief surface sharpen, and watch agreement lose weight when it is not independent.
Loading evidence laboratory
Architecture
Language is the interface.Mathematics is the engine.
A language model predicts plausible text. A decision under uncertainty needs models of time, dependence, volatility and propagation whose evidence can be challenged. Select a layer.
Loading architecture
Research
The sciencebehind the decision.
Research is infrastructure at Grand Thera: the method every inference in the stack must answer to, published openly so it can be challenged.
Working Paper v1.1 · Preprint · Conceptual framework
Published· Sep 2026Beyond Ontology
Computational Epistemology as a Layered Inference Machine
Fernando Negrini (lead author) · Hugo Dias · Darlan Parucker Lueders · Caroline Assis
Ontological coherence does not determine how a machine should infer under uncertainty. The paper introduces an applied computational epistemology: an open, layered inference architecture built above a computational ontology. Candidate signals are interrogated by heterogeneous engine families, subjected to repeated evaluation across samples, windows, frequencies, specifications and regimes, qualified, reconciled through dependence-aware convergence, calibrated into an event probability and mapped to business impact as an auditable artifact. A human modeler retains governance, and added complexity is admitted only when it yields statistically and economically material gains over the simplest adequate baseline.
The inference path · L0–L9
- L0Ontological data foundation
- L1Domain and decision framing
- L2Domain-specific inference engines
- L3Multi-scale robustness & persistence
- L4Evidence qualification
- L5Cross-model evidence synthesis
- L6Probabilistic synthesis, stress & fuzzy mapping
- L7Decision-impact mapping
- L8Decision-grade inference artifact
- L9Learning, recalibration & modeler governance
Newsroom
Thinking in public.Press and research notes.
In the press
(opens in a new tab)
PulseBrand · Opinion by Fernando Negrini · Sep 25, 2026
A high-stakes decision isn't solved by one more dashboard
For boards and CFOs, the relevant question is not how much information a company produces, but how much decision it can sustain.
(opens in a new tab)PulseBrand · Sep 1, 2026
Enterprise AI has become a commodity in automation. The next frontier is separating signal from noise
Most companies already capture efficiency gains from AI, but few grow revenue. Grand Thera bets that the missing layer is not organizing data but inferring which of it is evidence.
(opens in a new tab)
Blue Studio · PulseBrand · Sep 1, 2026
After the dashboard: what boards now demand from a decision system
Dashboards organize the past. For high-stakes decisions on financial positions, supply chains and exposure, boards now ask for a decision system.
(opens in a new tab)PulseBrand · Aug 17, 2026
Grand Thera announces a heavyweight advisory board with Guga Stocco and Rafael Navarro
The co-founder of Banco Original and Brazil's representative at CERN join Grand Thera's advisory board.
(opens in a new tab)PulseBrand · Jul 30, 2026
Artificial intelligence becomes the bet to unlock investor entry into tokenized assets
Ahead of launching its tokenized-asset platform, Guga Stocco's RedMind turns to Grand Thera's algorithms to tackle volatility and counterparty risk.
(opens in a new tab)
PEGN · PulseBrand · Jul 9, 2026
Not every tech company is a startup: multidisciplinary partners who became an AI deep-tech
A multidisciplinary trio that spent two years building before selling. Grand Thera's bet was born from a real pain, not from AI hype.
(opens in a new tab)PulseBrand · Jul 2, 2026
A deep-tech from Santa Catarina bets that the next cycle of AI is decision-making
After the generative-AI hype, the center of gravity shifts from content generation to decision-making. Grand Thera wants to define the category in Brazil.
Insights on Medium
All articles ↗
Institutional record
Grand Thera in numbers
Company-reported- Workloads deployed
- 750+
- Decision-critical use cases
- 20+
- Customers expanded
- 90%
- Events processed
- 10M
Founders
Mathematics, engineeringand governance.
Background in pure mathematics and physics, with more than twelve years across AI/ML and quantitative finance. Leads strategic vision, uniting science, business and disciplined execution.
Quant developer and AI/ML engineer with more than ten years engineering high-performance, mission-critical systems. Designs scalable architectures that fuse advanced software engineering with resilient infrastructure.
Law graduate with an M.Sc. in Business and more than twenty years in finance and advisory. Drives governance, board advisory and strategic partnerships to ensure accountability and alignment.