Agentic HR vs. Workday Sana

Track: 04 Glossary and Reference | Module: 04.2 Competitive Reference Cards
Reading time: ~6 min | Type: Competitive Card
URL: /academy/agentic-hr-vs-workday-sana/

---

The Core Question

Workday is the system of record for millions of employees worldwide. With the Sana partnership, Workday now has AI agent capabilities layered onto that record. Gloat’s Loomra engine powers a system of action that reads from Workday, SuccessFactors, Oracle, and dozens of other systems simultaneously, then drives outcomes across all of them.

The question is whether a smarter system of record is the same as a system of action.

This reference card compares the two approaches across six dimensions so you can evaluate where each delivers value and how they work together.

---

1. Deployment Model

Workday Sana

Workday’s partnership with Sana brings conversational AI and learning-oriented agent capabilities into the Workday ecosystem. The primary interaction surface remains the Workday portal. Sana agents handle employee self-service queries, surface learning content, and assist with Workday-native tasks like submitting requisitions or navigating manager workflows.

Workday has been improving its Microsoft Teams integrations, but the core experience is still portal-first. Employees must either open Workday directly or rely on limited push notifications to engage with Sana-powered features.

Gloat (Loomra)

Gloat delivers 29 agents directly into the tools where employees already work: Microsoft Teams, Slack, Google Chat, and Microsoft Copilot. There is no portal to visit. A manager gets a redeployment recommendation in Slack. A recruiter sees a calibrated candidate list in Teams. An HRBP receives a flight-risk alert in Google Chat.

This flow-of-work delivery model eliminates the adoption gap that portal-centric tools face. When the agent meets you where you are, usage is not optional; it is ambient.

---

2. Process Architecture

Workday Sana

Workday processes are governed by the Business Process Framework (BPF). Every workflow follows a predefined sequence of steps: approvals, notifications, condition rules, and to-do tasks. Sana can make these steps easier to navigate through conversational guidance, but it does not change the underlying sequential architecture. If your BPF requires five approvals, the AI still waits for five approvals.

This is a strength for compliance-heavy processes where deterministic sequencing matters. It is a limitation when the optimal next step depends on dynamic context that BPF cannot evaluate.

Gloat (Loomra)

Loomra’s Business Logic Engine processes 24 categories of governance rules in under 15 milliseconds. Critically, it extracts business rules from Workday and other systems, then enforces them dynamically. An agent does not follow a static flowchart. It evaluates the current state of the employee, the role, the org structure, and the applicable policies, then determines the right action.

This means the same agent can handle a lateral move in one business unit with two approvals and a cross-functional redeployment in another business unit with four approvals and a compensation review, without anyone configuring separate BPF sequences for each scenario.

---

3. Data Boundaries

Workday Sana

Sana operates within the Workday data perimeter. It can reason across Workday HCM, Workday Recruiting, Workday Learning, Workday Adaptive Planning, and other Workday modules. Within that perimeter, it has deep access and can surface connections that a human might miss.

The limitation is architectural. Sana cannot natively reason across your external ATS, your third-party LMS, your collaboration tools, your HRIS in a recently acquired subsidiary running SAP, or your labor-market data feeds. If the answer requires data from outside Workday, the agent cannot reach it.

Gloat (Loomra)

Loomra’s Knowledge Graph contains 2.4 million entities and 18.7 million edges, assembled from every connected system. It reads from Workday, SuccessFactors, Oracle, ATS platforms, LMS platforms, collaboration tools, and external labor-market sources simultaneously. Queries resolve in under 50 milliseconds.

When an agent evaluates whether to recommend an internal candidate for an open role, it does not just check Workday. It checks the candidate’s project history from Jira, their learning completions from Cornerstone, their peer endorsements from the collaboration layer, and the market salary benchmark for the target role. All in one pass.

And it writes back. When a decision is made, Loomra can update the Workday record, close the ATS requisition, and trigger the onboarding workflow in the appropriate system. Cross-system intelligence paired with cross-system execution.

---

4. Skills Intelligence

Workday Sana

Workday Skills Cloud is a taxonomy-based system. It maintains a curated library of skills with defined relationships. Employees and managers tag skills from the library. Machine learning suggests additional skills based on role and peer patterns. Skills Cloud works well within Workday’s own modules, powering skills-based routing for learning recommendations and talent reviews.

The taxonomy approach has known trade-offs. New skills must be added to the library before they can be recognized. Equivalencies between similar skills (e.g., “machine learning” and “ML engineering”) depend on manual curation. Cross-language skill matching is constrained by the taxonomy’s coverage in each locale.

Gloat (Loomra)

Loomra uses six specialized embedding models, including gloat-embed-entity-v3, achieving 90% accuracy in skill matching. Rather than mapping to a fixed taxonomy, the system understands skills semantically. It knows that “machine learning,” “ML engineering,” and the Mandarin equivalent refer to overlapping capabilities without anyone configuring that relationship.

The Personalization Engine draws on 5 million employee career patterns and over 1 billion career events collected across 8+ years of ML research. The 14 Intelligent Tools across 5 categories have processed over 200 million matches, each responding in under 100 milliseconds.

This is not a taxonomy with AI on top. It is an inference engine that understands what skills mean, how they relate, and how they evolve over time.

---

5. AI Architecture

Workday Sana

Sana brings its own AI models to the partnership, focused on conversational understanding and learning content delivery. The specifics of Workday’s model strategy are evolving, but the architecture is tightly coupled to the Workday platform. Customers adopt the AI that Workday and Sana provide.

Gloat (Loomra)

Loomra is model-agnostic. It integrates with Anthropic, Google, and IBM WatsonX, selecting the optimal model for each task. A summarization task might use one provider while a complex reasoning chain uses another. Customers with specific model requirements, whether for regulatory, sovereignty, or performance reasons, can configure their preferred providers.

The 29 agents span four categories: Talent Management (7 agents), Talent Acquisition (6 agents), Skills Intelligence (8 agents), and Workforce Planning (8 agents). Each agent orchestrates across the 14 tools, the Knowledge Graph, and the governance engine to complete end-to-end workflows, not just answer questions.

---

6. Coexistence Model

This is the dimension that matters most in practice. Most enterprises running Workday are not going to stop running Workday. The question is what layer sits on top.

Gloat treats Workday as the system of record and layers cross-system intelligence and action on top. Loomra reads Workday data, respects Workday business rules (extracting and enforcing them through the Business Logic Engine), and writes decisions back to Workday when appropriate.

This is not a replacement pitch. It is an architecture where Workday continues to do what it does well, managing the canonical employee record, running payroll, handling compliance workflows, while Gloat handles what Workday cannot: reasoning across system boundaries, delivering agents in the flow of work, and driving action based on signals that no single system can see alone.

Workday Sana makes Workday smarter. Gloat makes your entire HR technology stack smarter.

---

Summary Comparison

Dimension	Workday Sana	Gloat (Loomra)
Delivery surface	Workday portal (improving Teams support)	Teams, Slack, Google Chat, Copilot
Process architecture	Business Process Framework (sequential)	Dynamic orchestration, 24 governance rule categories, <15ms
Data boundaries	Workday-native modules	Cross-system: Workday + SuccessFactors + Oracle + ATS + LMS + market data
Skills approach	Skills Cloud (taxonomy-based)	6 embedding models, 90% accuracy, semantic matching
AI models	Sana + Workday proprietary	Model-agnostic: Anthropic, Google, IBM WatsonX
Knowledge base	Workday tenant data	2.4M entities, 18.7M edges, 1B+ career events
Agent count	Expanding (Workday-scoped)	29 agents across 4 categories
Tool ecosystem	Workday-native actions	14 tools, 5 categories, 200M+ matches
Write-back	Workday-native only	Cross-system (including back to Workday)
Coexistence	Workday-only	Additive layer on top of any SoR
Scale	Workday customer base	100+ enterprises, 5M+ employees, 112 countries

---

When to Choose What

Workday Sana is the right choice when:
– Your HR technology stack is entirely Workday end to end
– Your primary need is making Workday-native workflows easier to navigate
– You want conversational self-service for Workday transactions
– Your skills strategy is fully managed within Skills Cloud

Gloat is the right choice when:
– You run multiple HR systems (Workday + acquired entities on SAP, Oracle, etc.)
– You need agents that act in Teams, Slack, or Google Chat, not just in a portal
– Your skills intelligence must span internal and external labor-market data
– You want to preserve Workday as your SoR while adding a cross-system action layer
– You need model-agnostic AI with governance controls across 24 rule categories

Both together is the right choice when:
– You want Workday to remain your system of record
– You want Gloat to serve as the cross-system intelligence and action layer
– You need agents that can read from Workday, reason across all your systems, and write back

Most enterprise customers end up here. Workday handles the record. Gloat handles the action.

---