An AI agents course for operations, HR, sales, IT and training staff who want to build artificial intelligence agents that answer with their company's documents and execute tasks, without writing code. In 16 hours each participant builds a real agent for their area with no-code tools (n8n, Copilot Studio and models from OpenAI, Anthropic or Google), connects it to WhatsApp, Teams or email and leaves it measured and documented. DC-3 certificate filed with Mexico's STPS and Geedle Bot to review every step.
It is a 16-hour, hands-on, no-code program to build AI agents that work with your company's information: an agent that answers internal policy questions citing the document, one that qualifies and follows up leads on WhatsApp, one that solves questions about a course or a technical manual. It covers how to choose the use case, prepare the documents (RAG), assemble the agent flow with visual tools such as n8n or Copilot Studio, give it memory and tools, connect it to the channels where your people already are, test it with real questions and measure its accuracy. It runs on-site at your company, live virtual or self-paced on the platform, with a DC-3 certificate.
The exact duration is in the first card. It runs on-site at your company, live virtual for teams in several locations or self-paced on the platform; every participant who meets attendance and assessment receives a DC-3 certificate with the STPS, and Geedle Bot answers questions afterwards with the exact minute of the recording.
16 hours
Live virtual · On-site · Self-paced
DC-3 · STPS
Geedle Bot answers questions 24/7 with the exact minute of the recording
Proposal with three scope options and a closed price in under 24 business hours
Learning objectives observable on the job.
Identify each area's repetitive questions and tasks, estimate the savings and discard the agents nobody would use.
Organize manuals, policies and catalogs, remove old versions, chunk them and load them into a knowledge base the agent cites.
Assemble the full flow in n8n or Copilot Studio: trigger, model, memory, tools (search, send, log) and answers with sources.
Publish the agent on WhatsApp, Teams or email, test it with 20 real questions, measure accuracy and escalate to a person when it does not know.
Six modules in order, each with a deliverable you apply in your operation the following week. Exercises use real files and cases from your organization, each module's assessment is recorded on the platform and serves as evidence for the DC-3 certificate.
Agent, assistant and chatbot; language models, context and tools; risks (hallucinations, sensitive data) and how to bound them.
Use-case map for the areaInterview with the participant's own area, list of repeated questions, savings estimate and document preparation (cleanup, versions, permissions).
Loaded knowledge basen8n flow: trigger, chunk retrieval, model, answer citing the document; test with 10 questions.
Agent that answers with sourcesPer-conversation memory, calls to spreadsheets, CRM or calendar, forms and logging of every interaction.
Agent that executes tasksPublishing on the channel where the user is, identifying the person, hours and escalation to a human.
Agent published on a real channelBank of 20 real questions, accuracy and coverage, agent technical sheet, minimum AI usage policy and maintenance plan.
Technical sheet and 90-day planOperations, HR, sales, customer service, training and IT coordinators who want to build artificial intelligence agents without depending on a developer, and technical teams that prefer visual tools before coding. No programming required: the work is done with visual interfaces and templates. If your team does code, the AI for developers course is the next step.
Each participant leaves with a working AI agent for their area, connected to a real channel (WhatsApp, Teams, email or an internal page), with their documents loaded, 20 test questions with measured accuracy and a one-page technical sheet (what it does, what it does not, which data it uses, who maintains it). The company gains a short list of agents worth building, a minimum AI usage policy and people able to maintain and improve what they built, without waiting for a vendor.
Not a sliofck: a working, measured and documented agent, built by your own people.
An agent published on a real channel
By the end, each participant's agent answers on WhatsApp, Teams, email or an internal page, with their area's documents loaded.
Accuracy measured with 20 real questions
Each agent is tested with a bank of questions from the area and leaves with its accuracy rate and its list of improvements.
Technical sheet and usage policy
One page per agent (what it does, which data it uses, who maintains it) and a minimum generative AI policy for the company.
Geedle Bot to review every step
Our own agent, trained on the course, answers questions and takes you to the exact minute of the recording where each node was built.
Request the proposal for the ai agents course: three scope options with a closed price in under 24 business hours.
Operations, HR, sales and customer service are the areas with the most repeated questions and the most unused documents; an AI agent built by the area itself solves them and stays maintained in-house. If you would rather have us build it, see AI agents for business.
Student services, coordinators and libraries answer the same questions every semester; the course leaves agents that reply with the regulations and the calendar. The next step is the AI tutor for universities.
Agencies with procedures and frequent questions build agents that guide citizens with current regulations, with a DC-3 certificate per public servant. It is within the courses for government and the public sector.
All three get called "the bot" at the office, but they do not do the same thing. This table clarifies what you learn to build and when each one makes sense.
| Capability | Menu chatbot | Assistant with RAG | AI agent |
|---|---|---|---|
| How it answers | Fixed options and scripts | Writes with your documents and cites the source | Writes, cites and decides which tool to use |
| Sources | The ones you programmed | Updatable knowledge base | Knowledge base + systems (CRM, sheets, calendar) |
| Actions | None or links | None: it only informs | Logs, sends, schedules, looks up and escalates |
| Memory | No | Per conversation | Per conversation and per person |
| When it makes sense | 3-step procedures | Questions about documents | Processes with decisions and follow-up |
| You build it in the course | Module 1 (to tell it apart) | Module 3 | Modules 4 to 6 |
Real projects told with facts, not promises.
Train crews spread across restaurants in several Mexican states and the United States with the same standard — and give every employee a clear view of how to grow inside the company.
Read the full caseBring an operation that lives on the road under a quality management system that meets the new ISO 9001:2026 and passes the certification audit.
Read the full case
Geedle Bot is trained on the syllabus, materials and recordings of your course. Participants ask in plain language and get an exact answer, not a generic one.
On-site at your company in the nine locations where we have a team (Mexico City, Puebla, León, Guanajuato, Monterrey, Nuevo León, Guadalajara, Querétaro and Toluca), live virtual for teams in any city and self-paced on the platform across Mexico. Each location page explains the approach for the local industry.
Short reads to decide better before you hire.
Other applied artificial intelligence courses and the full category.
Tell us how many people, in which city and in which format. We reply the same day, free and with no commitment.
