Applied intelligence
Illustrative scenarios — not case studies with invented numbers. Each one shows the problem, what Yan.m builds and the path the data takes to the agent.
Industry. Intelligent industry
The problemThe operation depends on people cross-checking the ERP, spreadsheets and documents to decide what to produce, buy or stop. The information exists, but it is scattered, and the repeated task delays the decision.
What is built
The agent reads what the ERP, the database and the documents already record. It does not replace the factory floor: it prepares the context, runs the query and returns the next step. A person steps in when a case falls outside the rule.
What changes in the operation
- R1Less time gathering data from different systems
- R2Repeated tasks leave the manual queue
- R3Decisions made on the same facts for the whole team
- 01ERP and documents
- 02MCP server
- 03Operations agent
- 04Flow automation
Retail. Connected commerce
The problemService is split between WhatsApp, the CRM and the catalogue. Each channel answers differently, the history does not follow the customer and the conversation dies before the order.
What is built
A sales agent sits on WhatsApp with access to the CRM and the catalogue through MCP. It recognises the customer, checks availability and follows the store's script. The salesperson takes over when there is a negotiation, an exception or an unusual order.
What changes in the operation
- R1One history instead of loose conversations
- R2Answers aligned with stock and store policy
- R3The human team on the conversations that need judgement
- 01WhatsApp and CRM
- 02Catalogue via MCP
- 03Sales agent
- 04Continuous service
Services. Digital company
The problemThe company grows at the same pace as always: more clients means more people repeating scheduling, proposals and follow-up. There is no system holding the process.
What is built
The platform becomes the place where the work happens. Agents and automations handle what is the same in every contract — opening a record, chasing a step, updating a status — and leave the team with the relationship and the exceptions. MCP connects the platform to the CRM, the calendar and the documents.
What changes in the operation
- R1Growth without copying the same task for every new client
- R2Status visible to the whole team
- R3A base ready for the next digital product
- 01Own platform
- 02CRM and calendar
- 03Agents via MCP
- 04Follow-up automation
Education. Connected learning
The problemContent lives in the LMS, in files and in messages. Whoever follows the learning path does it by hand, and the student cannot find the next material without asking.
What is built
The learning agent queries the LMS and the content through MCP. It organises what has been covered, points out what is missing and keeps the follow-up going without someone rereading every track. The educator steps in to adjust the path, not to look for files.
What changes in the operation
- R1Material found in the flow, not in loose folders
- R2Follow-up that does not depend on a spreadsheet
- R3A path adjusted per person, on the same base
- 01LMS and content
- 02MCP server
- 03Learning agent
- 04Continuous follow-up
Next step
Shall we get this built?
Send the brief by email, WhatsApp or the form. I reply with the missing questions and then with scope, timeline and price in writing — no mandatory meetings.
- 01Written brief
- 02Fixed proposal
- 03Deliveries with online previews