GUÍA PRÁCTICA · CLOSELLY MCP
Tu formación,
en tu agente de IA
Seven practical questions to compare courses, focus follow-up, and prepare better decisions using authorized Closelly data from your own AI agent.
60 segundos · Demostración recreada con datos ficticios. Las recomendaciones mostradas no ejecutan acciones automáticamente.
Descripción y transcripción del video
Descripción Duración: 60 segundos. Audio en español y subtítulos integrados. Demostración con datos ficticios. El cuerpo del video es una recreación de ChatGPT conectado a Closelly mediante MCP: no es una grabación real ni muestra datos de clientes. Se comparan cursos ficticios con 80 %, 70 % y 38 % de finalización. En Venta consultiva, 24 de los 62 alumnos pendientes llevan 7 días o más sin conectarse a la plataforma: 10 sin iniciar y 14 en curso. El plan de seguimiento es sugerido; el asistente no ejecuta acciones. La portada y el cierre nombran ChatGPT, Claude, Gemini CLI, Copilot Studio y otros agentes de IA compatibles con MCP, según sus capacidades, permisos y configuración. Transcripción Tu formación. En tu ChatGPT. Conecta Closelly mediante MCP y consulta los datos autorizados de tu organización, desde tu propio asistente. Pregunta qué curso necesita atención. En este ejemplo, Inducción alcanza ochenta por ciento; Seguridad, setenta; y Venta consultiva, treinta y ocho. ChatGPT convierte los datos en una comparación clara. Profundiza: veinticuatro alumnos de Venta consultiva tienen el curso pendiente y llevan siete días o más sin conectarse a la plataforma. Diez no han iniciado y catorce están en curso. Son veinticuatro de los sesenta y dos pendientes. Prepara el seguimiento: contacta a esos alumnos, revisa sus barreras y vuelve a medir en una semana. Los datos orientan; tú decides. Closelly, conectado a ChatGPT y otros asistentes compatibles con MCP. Tus datos de formación, en tu conversación. Pide una demo.
Abrir el video MP4Compara cursos
Orienta la primera revisión con un criterio visible.
Enfoca el seguimiento
Distingue quién necesita ayuda para empezar o continuar.
Prepara decisiones
Llega a la reunión con hallazgos y preguntas por validar.
There are 62 people who have not yet completed a course. Who should receive support first? Those who never started? Those who made progress and then stopped logging in? What does their manager need to know to help?
That chain of questions comes up in any training team. Closelly MCP lets you query authorized courses, learners, progress, and reports from your own AI agent. ChatGPT, Claude, Gemini CLI, and Microsoft Copilot Studio support MCP; connecting each to Closelly depends on its capabilities, configuration, and enabled permissions. This article's demo uses ChatGPT as an example.
MCP stands for Model Context Protocol: a standard for connecting AI assistants with tools and data sources. In practice, Closelly provides the available information; the assistant helps organize it, explain it, and continue the analysis. The responsible person retains the decision.
Fictional data. Let's look at seven situations. All figures in this article are fictional and are used to show how a conversation might work. The three example courses each have 100 enrollments and the same delivery window. The initial snapshot is dated October 15, 2026.
Choose which course to review first
Question: “Compare completion across my courses. Include both the percentage and number of completed enrollments. Where should I look more closely?”
Illustrative response: “Onboarding: 80 out of 100, or 80%. Operational Safety: 70 out of 100, or 70%. Consultative Selling: 38 out of 100, or 38%. The last course has the lowest completion.”
CRITERIO PARA DECIDIR
The benefit is guiding the first review with a visible criterion. The assistant can present a table or chart if the environment supports it. Before prioritizing, the responsible person checks dates, duration, and requirements: a newly opened course may have lower completion without performing poorly. The comparison opens a specific investigation.
Help people start or continue
Question: “In Consultative Selling, separate completed, in-progress, and not-started enrollments.”
Illustrative response: “38 completed, 42 in progress, and 20 not started. There are 62 outstanding.”
CRITERIO PARA DECIDIR
The team can now prepare different kinds of support. For the 20 people who have not started, check access, awareness of the invitation, and availability. For the 42 who have already begun, understand what they need to continue. Managers validate those needs before choosing the support. The benefit is avoiding a single reminder that treats different situations as though they were the same.
Define a manageable follow-up group
Question: “Among people with Consultative Selling outstanding, how many have not logged in for at least seven days? Separate them by status.”
Illustrative response: “24 of the 62 outstanding enrollments meet the criterion: 10 not started and 14 in progress.”
CRITERIO PARA DECIDIR
The question turns a broad population into a defined group to review this week. The criterion uses the last platform login, which must be distinguished from the last activity within the course. An unknown login date needs separate handling. The responsible person checks holidays, shifts, and access difficulties before making contact. The benefit is focusing follow-up without labeling anyone as disengaged or predicting dropout.
Compare groups with context
Question: “Within this course, compare the proportion of people who have not started in two authorized groups. Include numerator and denominator.”
Illustrative response, in an independent scenario: “Group A: 12 out of 20, or 60%. Group B: 12 out of 60, or 20%.”
CRITERIO PARA DECIDIR
Both have 12 people who have not started, but their relative scale differs. If the groups and their enrollments are available, filters can gather the counts and the assistant can calculate the proportions. The responsible person checks that the dates are comparable and reviews how the invitations were sent. The benefit is allocating attention with context; the numbers alone do not explain the difference or justify evaluating a manager.
Prepare a meeting that leads to decisions
Question: “Using these results, prepare a summary for the training meeting: three findings, two open questions, and one proposed next step.”
Illustrative response: “Consultative Selling has 38 completed enrollments out of 100. There are 62 outstanding, and 24 meet the follow-up criterion. Their barriers still need to be understood, and the availability of support confirmed. Proposal: review those cases and agree on an owner and a date for the next measurement.”
CRITERIO PARA DECIDIR
The assistant transforms the queried data into a usable draft. The benefit is dedicating the management conversation to priorities and responsibilities. The presenter validates the source, snapshot date, and figures, and decides which information is appropriate to share with that audience.
Prepare relevant messages
Question: “Suggest two short drafts: one for people who have not started and another for those in progress. Ask what support they need without assuming the reason.”
Illustrative response: “To get started: ‘Were you able to access the course? Let us know if you need help getting started.’ To continue: ‘Is anything making it difficult to move forward? We want to understand what support would help.’”
CRITERIO PARA DECIDIR
The benefit is moving from a finding to concrete preparation for support. The assistant drafts these texts using the available context; this query does not send messages from Closelly. The responsible person reviews the tone, channel, and recipients and authorizes any sending. Contact information is used only when the corresponding permission exists.
Measure what changed afterward
Question: “Compare the saved October 15 snapshot with the current query. Use the same course, the same cohort, and the same definition of completion.”
Illustrative response: “With the same 100 enrollments, completions rise from 38 to 50: 12 additional completions and an increase of 12 percentage points.”
CRITERIO PARA DECIDIR
The benefit is closing the follow-up cycle with evidence that can be reviewed. This requires retaining the earlier snapshot and querying the current one; the example does not assume MCP provides an automatic history. The responsible person checks additions, removals, and reassignments and documents the result. The observed change can help adjust the plan, although it does not by itself prove that an intervention caused the improvement.
Greater clarity, with a well-defined scope
These conversations depend on the tools, data, and permissions enabled in each organization. MCP Business queries the organization authorized in the session. Personal data and exports require specific permissions; a connection does not open access to the entire platform or to other companies.
Asking a well-defined question also matters: specify the course, group, snapshot date, and unit of measurement. One person can have multiple enrollments. Completion and average progress measure different things. Keeping those definitions visible makes the analysis more useful and easier to review.
Start with a specific question
Value emerges when one question leads to the next: which course to review, which group to support, what information is missing, and how to check later whether progress occurred. Closelly MCP helps bring that journey into a conversation with authorized data and explicit decisions.
Think about your next training meeting. Which answer would help you arrive with a clear priority? That is a good starting point for a Closelly Insights and MCP demonstration tailored to your organization.