GFoundry Intelligence: from a question to a useful next step
Creating a course, practising a difficult conversation and exploring training data require different kinds of help. GFoundry Intelligence brings that help into the platform through specialised agents and intelligence services, with a defined purpose for each.
TOAR — Talent Orchestration, Automation & Response — describes the talent management Framework: bring relevant data together, reduce repetitive work and turn evidence into a considered response. GFoundry Intelligence is the product offering that supports this Framework. The Framework structures how the capabilities are used; GFoundry Intelligence provides the agents and services.
The distinction matters in practice. An answer from an agent, a recorded training completion and an Insight on a dashboard are different kinds of evidence. Connecting them helps HR decide what to do next, while keeping the source, permissions and responsibility for the decision clear.
GFoundry
GFoundry Intelligence · TOAR Framework
Talent Orchestration, Automation & Response
TOAR Framework
A single layer over the systems you already have. Data stops living in silos and starts being read together.
Specialised agents help with defined tasks. People review content and approve the next step.
Available evidence helps identify where to investigate, support learning or prepare a development conversation.
Specialised agents and analytical Insights. Each service has its own inputs, permissions and limits.
the engine
The agents: a defined role for each kind of work
Gi provides the intelligence framework. The agents apply it to specific jobs, using the inputs and tools relevant to that job. A course briefing, a company policy and an analytics question are different inputs; they are not interchangeable.
Workplace Assistant · Gi Assist
Employees and managers ask everyday questions about company knowledge, policies, contacts and learning resources. Gi works from the documents and platform content made available by the organisation. Check available sources when the answer informs an important action.
Learning Designer · Gi Learn
A course starts with a PDF or a briefing. The Learning Designer clarifies the audience, objective, structure and source scope, then presents a summary for confirmation before generating. The result is a course to review, with a structured page and the chosen slides, quiz, PDF or curated video. A curated video is selected from existing material, not produced by the agent.
Role-Play Coach · Gi Practice
Practice is a separate job from course creation. The Role-Play Coach uses a scenario and an AI persona to rehearse conversations, by text or voice. Feedback and evaluation follow the scenario’s criteria. The result supports practice and coaching; it is not a verdict on an employee’s personality or suitability.
Talent Strategist · Gi Talent
Use a natural-language question to explore available talent data and prepare a development discussion. The conversational analytics capability, documented as the Data Analyst Agent, returns analyses, tables, charts and Excel exports within the user’s access scope. This analytical tool reads records; it does not change evaluations or make employment decisions. Start with a question such as “Which teams have mandatory training still to complete?” and verify the figures before acting.
AI Writing Assistant · support inside forms
The AI Writing Assistant helps write, improve, shorten, correct or translate text in backoffice forms. Its output remains editable. Finish and review the original content before translating it into the languages configured for the organisation, then check each version before saving.
People Intelligence and Talent Insights: evidence to read
Alongside the generative agents, People Intelligence provides a configurable organisation dashboard built from a catalogue of Insights. GFoundry defines the metrics; HR selects the cards. Each card states its meaning, source date and limitations. A chart from a conversational query is not automatically a catalogue Insight.
Talent Insights brings the individual view into the employee profile, supporting development conversations with the available evaluation, learning and talent context. These are analytical services: their core Insights are computed in the platform rather than invented by an LLM.
Read the classification of each Insight. A factual measure, a diagnostic reading, an experimental signal and a validated prediction carry different weight. Missing data must remain visible. The current responsible-use rules exclude individual exit-risk assessments and anonymous survey answers from profile cards. People decisions stay with accountable people.
Explore the product views: People Intelligence · Talent Insights.
A practical example: from a policy to learning and follow-up
Consider a new customer-service policy. The objective is not merely to distribute a PDF, but to help teams apply the policy consistently. The following is an example of work coordinated by the team, rather than an automatic chain triggered by one prompt:
- Define the source and outcome. Choose the approved policy, the audience and the behaviour the course should teach.
- Prepare the course. Brief Learning Designer, confirm its proposed structure and review the generated material before publishing.
- Explain and practise. Use Writing Assistant for the announcement and Role-Play Coach for a customer conversation with clear practice criteria.
- Answer and analyse. Workplace Assistant helps people find the available policy information. Use conversational analytics to check training completion.
- Follow up. Read the relevant Insights alongside team context, then agree support or development actions with managers.
This is the TOAR Framework in concrete terms: relevant information, assisted preparation and a response grounded in evidence. Each step retains its own review and permissions.
Privacy and responsibility depend on the service
Do not assume that every input is anonymised. The analytics agent substitutes employee identifiers in structured data; this is pseudonymisation, not complete anonymisation. Questions, uploaded documents and practice conversations can contain personal information and may be sent as written to an external AI provider. The data used by an Insight follows a different processing route from a generative request.
Use the necessary information for the task, review the GFoundry Intelligence privacy policy and confirm the applicable processing terms. AI can make mistakes. Review important answers, learning content and analyses, and keep decisions about people with the responsible team.
What the Framework changes
Many organisations already have the records they need, but those records are scattered across systems and teams. A training completion is in one place, feedback in another and development goals in a third. The TOAR Framework connects three practical questions: what information is relevant, which work can be assisted, and what follow-up is justified?
MCP: connecting external AI tools to GFoundry
MCP (Model Context Protocol) is an open standard for connecting AI applications to external tools and data. In GFoundry, it provides an integration route for compatible external AI tools. It is distinct from using the agents inside the platform.
With an authorised, configured connection, an external tool such as Microsoft Copilot, ChatGPT or Claude can access the GFoundry tools made available to that integration. Depending on the exposed capabilities and permissions, the workflow may include:
- query platform records and metrics;
- analyse data from separately authorised sources;
- prepare or create content through the available tools;
- execute permitted actions within the agreed integration scope.
MCP access is included in Premium. A client supporting the protocol does not automatically gain access to GFoundry or to every function. Other platforms require their own connections, and the scope and any external-service usage are agreed separately. See the current plans and Intelligence capabilities.
- Identity and scope — confirm the authorised user, available tools and permitted operations.
- Data sent — review the fields, documents and free text that each connected tool receives.
- Provider conditions — check the connected service’s processing, retention and data-use terms.
- Action controls — agree confirmations, review and logging appropriate to the workflow.
GFoundry
The MCP approach
External AI tools · authorised integration
with authorised access
ATS
Payroll
Time & attendance
Leave & absence management
Other data sources
The three layers of enterprise software — and where GFoundry sits
Enterprise software is organized today into three layers, and understanding this split helps place GFoundry. The System of Record keeps the authoritative truth — the organization’s reference data; it’s the memory. The System of Engagement is the interface where people work every day — the hands. And the System of Intelligence & Action helps interpret information and prepare or execute authorised tasks — the brain.
GFoundry spans all three layers. It is the System of Engagement for talent — where people learn, are recognized, set goals and communicate — and, with GFoundry Intelligence and configured MCP integrations, it is also the System of Intelligence & Action that supports analysis, preparation and authorised actions. But it is equally a System of Record for a significant part of people data: it’s in GFoundry that the records of onboarding, mandatory and optional training, skills and proficiency levels, performance reviews, OKRs, peer recognition, feedback and behavioral data (interactions, interests, profiles, metric evolution) are generated and stored. The “source of truth” isn’t only financial or legal — in the talent and development domain, the source of truth is GFoundry. Core HR/ERP keeps the HR records, payroll and financial and legal compliance, which GFoundry integrates with and orchestrates.
How GFoundry Intelligence fits your architecture
GFoundry complements the Core HR, HCM and ERP systems already in use. APIs and configured connectors handle data exchange; MCP exposes permitted tools to an external AI client. SSO and identity services such as SAML, Active Directory or LDAP address authentication, which is a different task. For organisations with several business units, the Multi-Company setup and available languages are defined for the project. Data consolidation and each unit’s experience depend on that configuration.
It’s a model designed above all for medium and large companies, where the complexity of data and processes justifies orchestration, and where adoption — not just functionality — determines the return.
Frequently asked questions
The agents, the TOAR Framework and the role of MCP.
What is the difference between TOAR and GFoundry Intelligence?
TOAR means Talent Orchestration, Automation and Response: a talent management Framework for organising data, assisting work and deciding the next step. GFoundry Intelligence is the product offering, with specialised agents and intelligence services. TOAR is not a separate product.
Which agents are available?
Workplace Assistant (Gi Assist), Learning Designer (Gi Learn), Role-Play Coach (Gi Practice) and Talent Strategist (Gi Talent) support different tasks. AI Writing Assistant helps inside backoffice forms. Availability depends on the plan, enabled capabilities and user permissions.
Does Learning Designer publish a course automatically?
It clarifies the briefing and asks for confirmation of the proposed course before generating. The generated course still needs review and editing before publication. Its chosen video is curated from existing material, not generated by the agent.
Are People Intelligence Insights generated by an LLM?
Their core calculations run in the platform. People Intelligence offers a defined catalogue for the organisation dashboard, while Talent Insights provides the individual profile view. These services are distinct from asking an agent for an additional analysis. Read the source date, classification and limits of each card.
What is GFoundry's MCP approach?
MCP connects a compatible external AI client to the GFoundry tools exposed by an authorised integration. The available queries and actions depend on configuration, permissions and agreed scope. MCP access is included in Premium and does not mean unrestricted access to the platform.
Which LLMs can I use with GFoundry?
Copilot, ChatGPT and Claude are examples of external AI tools. Confirm that the chosen client supports the required MCP connection and authentication, and that the GFoundry integration exposes the tools needed. Compatibility with the protocol alone does not configure or authorise the connection.
Does this replace my Core HR or HCM?
GFoundry holds the talent and development records created in its modules and complements your Core HR, HCM and ERP. Data exchange uses configured integrations; authentication uses identity services such as SSO. The scope and responsibilities of each system are agreed for the project.
Is all information sent to AI anonymised?
No. Structured analytics data uses substitute identifiers, which is pseudonymisation. Questions, document excerpts, uploaded files and practice conversations may contain personal information. Processing and provider conditions depend on the service and supplied content. Consult the GFoundry Intelligence privacy policy and applicable processing agreement.
How is privacy handled in a connected MCP workflow?
Review the authorised user, available tools, data sent and action controls. A connected external AI service has its own processing, retention and data-use conditions. Agree the necessary reviews and logging for the workflow instead of assuming every connection has identical privacy guarantees.
What is included in Plus and Premium?
Plus and Premium include Gi, AI Writing Assistant, Learning Designer and Role-Play Coach. Premium also includes Talent Strategist, People Intelligence, Talent Insights and MCP access. AI-agent usage consumes the available shared credits; viewing included dashboards and profile Insights does not. Check the current plans for scope and usage conditions.
Is this only for large companies, and how long does implementation take?
The Framework is particularly useful when records and processes are spread across teams or systems. The first project should agree its users, content, integrations and review owners. Its timetable depends on that scope and readiness. Enterprise plans start at 250 users; speak with GFoundry about smaller teams and the appropriate option.
Start with one defined use case
A useful first project starts with a concrete problem and a short diagnosis:
- Which systems hold the relevant records, and which connections are already authorised?
- Which manual task would benefit most from an agent or a configured workflow?
- Are the sources accurate, current and sufficient for the intended analysis?
- Who will review the output, approve the next step and measure the result?
The answers define the first use case and the pace of the rollout. Book a demo to explore GFoundry Intelligence in your organisation, or see the talent management platform and how to structure your HR tech stack.