Equities
Brinson-Fachler
Allocation · Equity Selection · Interaction
Privately Deployed AI for Asset Managers and Asset Owners
Hedge Funds · Pension Funds · Family Offices · Endowments · Insurance
Accelerate model fine-tuning
Illuminate where alpha is created, lost, and left on the table.
Measure what the research actually contributed
Separate analyst insight from portfolio implementation.
ZTA measures analyst paper-portfolio performance against approved benchmarks and compares it to actual portfolio outcomes.
Illustrative figures, not client results
Paper Portfolio Return8.7%
Benchmark Return6.9%
Active Return+1.8%
Implementation Shortfall-0.4%
Net Value Add+1.4%
The paper portfolio holds every recommendation, thesis, and committee decision — including the ideas that never left the page. Positions are read directly from your OMS, so nothing is added to the PM’s workflow. Implementation shortfall then separates what the research was worth from what the implementation returned, closing the loop between research and results.
Common metrics aggregate. Methodology-specific attribution effects stay separate — Brinson-Fachler for equities, Campisi for fixed income — with Apis, ZTA’s governed AI research assistant, explaining what drove the result.
Fixed Income
Campisi
Income · Treasury · Spread · Fixed Income Selection
From research to decision
Turn model output into governed institutional judgment.
ZTA connects research, model analysis, Human Judgment, recommendation management, and outcomes into one governed workflow.
ResearchApis / Model CouncilInsightsThesisRecommendationClearancePM DecisionPortfolio ActionOutcome
Insights are provisional.Model output remains separate from the analyst’s thesis until intentionally promoted.
Human Judgment is authoritative.Apis guides. The Model Council challenges. The investment professional decides.
Lineage survives the workflow.Evidence, models, policy, rationale, PM disposition, action, and outcome remain connected.
Research Management · Credit deterioration assessment
Assessment
Does recent financial and operating evidence warrant changing the internal credit view from Stable to Deteriorating?
Evidence set: 14 approved sources.
Model Council
Response A · 4.10
Evidence supports a cautious view but stops short of a rating change.
Response B · 4.42
Greater weight to deteriorating liquidity and covenant headroom, with a clearer downside-first framing.
Response C · 4.55
Evaluator recommendation. Strong on evidence coverage, less aligned to the firm’s risk posture.
Response D · 3.91
Underweights the change in operating flexibility and overstates the stabilizing factors.
Model Assurance | Evidence-Led Qualification
Generic benchmarks do not tell you which models are fit for your workflows.
ZTA Labs delivers Continuous AI Assurance through two complementary disciplines. Model Assurance evaluates individual models for accuracy, reliability, evidence fidelity, and operational suitability. Council Assurance tests the collective system for independent challenge, meaningful disagreement, evaluator reliability, and governance integrity. Together, they provide the evidence needed for qualification, ongoing monitoring, and informed Human Judgment.
Open-Weight Model Ecosystem
United States
- Nemotron
- Microsoft
Canada
France
UAE
China
Baseline Model Assurance2 weeks
Establish the evidence before production.
Evaluate candidate models against a governed, client-specific calibration corpus. Test answer quality, execution reliability, evidence fidelity, evaluator integrity, and the independent contributions models make to the Model Council.
- Model Assurance baseline and qualification findings
- Model-level strengths, weaknesses, and failure analysis
- Evaluator and evidence-integrity assessment
- Recommended Model Council composition
- Governance findings and readiness recommendations
Quarterly Model AssuranceEvery quarter
Keep model qualification current.
Re-evaluate approved and candidate models as versions, runtimes, configurations, and institutional requirements change. Test evaluator stability, detect performance regressions, reassess Council contribution, and incorporate recorded Human Judgment and outcomes where available to guide governance recommendations.
- Model and evaluator drift assessment
- New model and version qualification
- Reliability and evidence-fidelity testing
- Model Council composition review
- Documented requalification and governance recommendations
Sample Model Assurance Baseline
See the methodology in practice.
A governed model assessment with model-level findings, failure analysis, source-level auditability, governance requirements, and deployment recommendations.
In a historical ZTA Labs Council assessment, a model with no unique best-answer wins was reported to have broken false unanimity three times. The finding illustrates why Council Assurance examines independent challenge and model complementarity rather than relying solely on average benchmark rankings.
Historical research observation; not a current Council qualification or a measured human-decision outcome. Reported in the September 2026 ZTA Labs research compilation. The underlying seat-contribution calculation was not retained, so the count has not been independently reproduced.
Finding
Historical testing identified a contradiction between a benchmark answer key and its evaluation policy.
After correction
After the policy was corrected, the candidates still failed because their answers omitted material information disclosed in the source evidence.
Scope
Historical research observation from a September 2026 assessment. Not a current model qualification and not a production readiness result.
ZTA Labs evaluates both model behavior and the integrity of the instruments used to measure it.
Why the architecture has to change
AI adoption is moving faster than institutional control.
Asset managers are adopting AI across research and investment workflows. When model providers, governance controls, and decision records remain disconnected, institutions can lose visibility into how Human Judgment informs consequential decisions.
Control boundary
Proprietary research and institutional context can cross organizational boundaries as AI usage expands.
Auditability
Model output is often difficult to reproduce, challenge, and trace back to the evidence and controls that shaped it.
Institutional memory
AI-generated knowledge can remain trapped inside tools instead of becoming durable institutional intelligence.
Model dependency
Reliance on external providers can constrain flexibility, while changes in models, runtimes, and institutional requirements create a continuing need for evaluation.
The problem is not access to AI. It is building an operating environment the institution can govern.
Built for extension
A governed platform your team can build on.
ZTA provides the governed core. Your team builds the dashboards, agents, workflows, connectors, and applications that make it yours.
Your team builds
Institution-specific experiences and intelligence
Your dashboards
Consume governed ZTA data through APIs
Your agents
Build firm-specific intelligence
Your workflows
Encode the way your investment process works
Your connectors
Integrate institutional and external systems
- Research repositories
- Market data providers
- Portfolio systems (OMS/PMS)
- Alternative data
- Your internal systems
Your applications
Build experiences beyond the ZTA UI
- Custom tools
- Internal applications
- Reporting and analytics
- Client portals
- Third-party integrations
Published extension layer
Stable, secure interfaces for building on ZTA
REST APIs
Query and retrieve governed data
SDKs
Build agents and tools
Events
Real-time updates and webhooks
Connectors
Integrate your systems and data
Policy hooks
Extend governance within your workflows
ZTA governed core
Model-agnostic and built for the institution
Identity and Entitlements
Who can access what
Policy and Guardrails
What intelligence may do
Model Control
Approved, replaceable models
Evidence and Audit
Provenance and traceability
Decision Lineage
From evidence to judgment to decision
Attribution
Research to portfolio to return
Your IP above the line. ZTA governance below it.
- Fully implemented external APIs
- Permissioned and auditable
- Model-agnostic and flexible
- Deployed in your environment
Built on Open Source
- xgrammar
Built on measurable controls
Evidence before claims.
ZTA Labs is built around documented, versioned methodologies rather than vendor assertions or opaque AI scoring.
Client-specific Model Assurance
Models are tested against institutional workflows, not only generic benchmarks.
Evaluator assurance
ZTA tests the instrument used to judge models, not just the models themselves.
Versioned methodology
Model Assurance, attribution, and governance methodologies are designed to remain documented and auditable.
Human Judgment captured
The architecture is designed to record what investment professionals decide, their reasoning, and the evidence supporting those decisions.
Decision lineage
The architecture is designed to preserve traceability across evidence, model outputs, policies, Human Judgment, PM disposition, and recorded outcomes.
A practical path to production
Start small. Prove the architecture. Scale with evidence.
ZTA Labs engagements are designed to move from a bounded first workflow to a production environment with clear scope, timing, and commercial expectations.
Path 1
ZTA Architecture
Prove the full architecture on your data and workflows.
Path 2
Standalone Model Assurance
Benchmark models and establish a baseline.
Optional Standalone Model Assurance is available independently or alongside a ZTA Architecture trial.
Discover
Define scope, objectives, and success criteria.
30–60 minutesProve
One-week ZTA Architecture Trial on your data.
1 weekDesign
Production architecture, data, models, governance and integrations.
2–4 weeksBuild and Validate
Implement and validate the first governed workflow.
6–12 weeksOperate
In production inside your environment.
OngoingQuarterly Model Assurance
Reassess models, Council contribution, evidence, and drift.
Every quarter
Fixed pricing
Clear scope and commercial terms before deployment begins.
Tailored to your environment
Pricing depends on workflows, data sources, models, integrations, and governance requirements.
From pilot to production
Scale through evidence and institutional approval.
The Buzz · ZTA Labs Research
Thinking about what comes next in institutional AI.
Research and perspectives on models, architecture, governance, AI economics, Investment Research, and institutional intelligence.
Open-Weight AI Has Crossed the Enterprise Threshold
The capability gap has not disappeared. The deployment decision has changed.
Public evidence shows that open-weight AI has become a credible enterprise architecture option. But leaderboard performance cannot establish whether a model is fit for an institution’s work.
Earlier: The Seven Steps to AI Independence →
Start with one workflow
See whether ZTA fits your environment.
Define a specific investment workflow, data boundary, and governance requirement. ZTA Labs will help determine the right first step—Architecture Trial, Model Assurance, or both.