AI and Prompt Engineering – Advanced Level. AI Agents, Process Automation and Workflow Design

1. Advanced use of language models in complex, analytical, and process work

• selection of the AI model for the type of task,
• working with long context and large sets of information,
• comparing results generated by different models,
• planning a multi-stage way of solving a problem,
• limiting errors resulting from incomplete context,
• building repeatable patterns of working with language models.

2. Designing complex prompts based on roles, rules, input data and the expected result

• building prompts system and operational,
• defining the role, goal and scope of the model's responsibility,
• creating precise andstructions and constraints,
• using examples as response patterns,
• using reusable prompt templates,
• designing response formats for reports, analyses and recommendations,
• separating input data from andstructions.

3. Creating multi-step AI workflows using successive prompts, intermediate results and validation mechanisms

• dividing processes into specialized stages,
• passing results between successive steps,
• prompt chaining,
• creating sequences of analysis, evaluation and recommendations,
• building checkpoints,
• validation of intermediate results,
• repeating selected stages after detecting errors,
• designing repeatable workflow business processes.

4. Advanced analysis of documents and large sets of information using language models

• analysis of many documents simultaneously,
• comparing content a and identifying discrepancies,
• data extraction according to an established scheme,
• classifying a organizing information,
• generating summaries a reports,
• identification of missing data,
• formulating recommendations based on documentation,
• creating mechanisms for verifying sources a responses.

5. Advanced use of AI for data analysis, generating conclusions and supporting business decisions

• building analytical queries for data,
• interpreting tables i statements,
• detecting trends, deviations i relationships,
• generating business hypotheses,
• comparing scenarios,
• preparing decision recommendations,
• creating management reports,
• controlling the correctness of conclusions generated by AI.

6. Designing specialized AI assistants and agents carrying out specific business tasks

• differences between an assistant, an agent and an automatic workflow,
• defining the goal and scope of the agent's operation,
• designing and system instructions,
• determining available sources of knowledge,
• defining action-taking rules,
• building analytical, documentation and process agents,
• designing human acceptance points,
• limiting the scope of the agent's autonomy.

7. Designing agent workflows involving analysis, decision-making, tool use and execution of subsequent actions

• architecture of the agent process,
• planning actions by the agent,
• division of tasks into subsequent stages,
• use of documents and external data,
• performing actions based on the results of the analysis,
• handling conditions and exceptions,
• escalation of decisions to the user,
• control of the process flow,
• reporting of completed actions.

8. Integration of AI with documents, applications and data sources used in the organization's daily processes

• designing the flow of information between AI and applications,
• working with company documents,
• using knowledge bases,
• integration with forms and processes,
• transferring data between tools,
• creating automation scenarios,
• updating information used by AI,
• controlling permissions to data and tools.

9. Multi-Agent Workflow and designing the collaboration of multiple specialized agents in one business process

• designing the division of roles between agents,
• specialization of agents according to the type of task,
• transferring results between agents,
• coordinating the sequence of actions,
• supervising agent and executive agents,
• parallel execution of selected stages,
• resolving conflicts between results,
• mechanisms for final validation of results,
• examples of applications in analysis, reporting and process handling.

10. Creating advanced standards and prompt libraries supporting the repeatable work of entire teams

• designing company prompt templates,
• standardizing the structure of commands,
• creating prompt libraries according to processes and departments,
• versioning and updating prompts,
• documenting the method of using prompts,
• testing the effectiveness of various variants,
• building criteria for the quality of responses,
• preparing prompts for use by various users.

11. Quality control, security and governance in advanced AI-based solutions

• identification of risk in processes supported by AI,
• protection of personal data i confidential information,
• limiting hallucinations i incorrect recommendations,
• designing Human-in-the-loop mechanisms,
• control of access to data i tools,
• audit of actions performed by agents,
• monitoring of the quality of results,
• responsibility for decisions supported by AI,
• principles of secure implementation of automation.

12. Design workshop covering the creation of a complete AI process from task analysis to a ready workflow or agent

• selection of a business process for improvement,
• analysis of the current method of process execution,
• identification of stages that can be supported by AI,
• designing the architecture of the workflow,
• preparation of prompts i instructions,
• definition of input data i results,
• designing validation mechanisms,
• testing the prepared solution,
• optimization of process operation,
• preparation of a plan for implementing the solution in the organization.