Agentic AI & Workflow Automation

What you will be able to do

• Explain the difference between deterministic workflows and agents, and choose the simpler option that meets the need
• Define tools for a model with clear schemas, and validate every argument before executing it
• Build common workflow patterns: chaining, routing, parallel steps and evaluate-and-revise loops
• Build an agent loop with step, time and cost limits and a clear stopping condition
• Integrate business systems through REST APIs, webhooks and scheduled jobs with retries and idempotency
• Design human approval gates and audit trails for consequential actions
• Test automations with unit tests, mocked model responses and scenario evaluations
• Monitor automations in production with traces, usage and cost tracking, alerts and a way to pause them
• Apply privacy, consent and least-privilege practices to automations that handle personal data

About this course

Syllabus

This syllabus is a draft awaiting instructor review. Topics, order and schedule may change before the first batch is announced.

1. 01Tool calling foundations+3 live sessions · 6 hours · What automation, workflows and agents are, and how a model proposes tool calls that your code validates and executes.

   Learning objectives

   • Distinguish rule-based automation, LLM workflows and agents, and recognise when AI adds nothing
   • Define tools with clear names, descriptions and argument schemas
   • Validate tool arguments and handle tool errors without letting the model act unchecked
   • Break a business process into steps and identify which ones are safe to automate

   Live sessions

   1. 01Automation, workflows and agents2 hCourse setup; rule-based automation versus LLM workflows versus agents; mapping a business process into steps; where AI helps, where it adds risk and where a simple script is better; the course project brief.
   2. 02Tool calling fundamentals2 hHow tool calling works: the model proposes, your code decides and executes; tool schemas and descriptions; validating arguments with Pydantic; returning results and errors to the model; parallel tool calls; read-only versus side-effecting tools.
   3. 03Tool calling lab and doubt-clearing2 hBuilding lookup tools for the course project’s records, testing how tool descriptions change model behaviour, handling malformed arguments; open questions.

   Practice

   Map the enquiry-handling process into steps, then build three read-only tools (record lookup, course search and FAQ search) that a model can call with validated arguments.

   Assessment

   Quiz on workflows, agents and tool calling, plus a short process-mapping assignment reviewed with feedback.

   You will be able to

   You can give a model well-defined tools and keep your code in control of what actually runs.

2. 02Workflows versus agents+3 live sessions · 6 hours · Build predictable workflow patterns first, then agent loops for tasks that genuinely need flexible decisions — with limits that keep them in check.

   Learning objectives

   • Build chaining, routing, parallel and evaluate-and-revise workflows
   • Build an agent loop with planning, memory and a clear stopping condition
   • Set step, time and cost limits, and handle loops that fail to finish
   • Justify the choice between a workflow and an agent for a given task

   Live sessions

   1. 04Workflow patterns2 hPrompt chaining with checks between steps, routing by classification, running steps in parallel, evaluate-and-revise loops, keeping state between steps, when a fixed workflow is the right answer.
   2. 05Agent loops, memory and limits2 hThe observe-decide-act loop, planning, short-term and stored memory, stopping conditions, maximum steps, timeouts and cost budgets, detecting repeated or stuck behaviour, frameworks versus plain code.
   3. 06Pattern lab and doubt-clearing2 hBuilding the same task as a workflow and as an agent, then comparing reliability, cost and latency on the same inputs; open questions.

   Practice

   Build the enquiry classifier and reply drafter as a routed workflow, then build an agent version and record which performs better on twenty sample enquiries and why.

   Assessment

   Assignment: both implementations with a short comparison of reliability, cost and latency, and a recommendation.

   You will be able to

   You can choose and build the simplest pattern that solves a task, and keep agents bounded when you use them.

3. 03API integrations and reliable automation+3 live sessions · 6 hours · Connect automations to real business systems through APIs, webhooks and schedules, so that they keep working when networks and services fail.

   Learning objectives

   • Integrate REST APIs with API keys and OAuth, pagination and rate limits
   • Trigger automations from webhooks and schedules and verify incoming events
   • Make side-effecting steps idempotent and retry failures safely
   • Expose tools through the Model Context Protocol and give each tool least-privilege access

   Live sessions

   1. 07Integrating business APIs and webhooks2 hREST APIs, API keys and OAuth scopes, pagination and rate limits, receiving webhooks and verifying signatures, scheduled jobs, connecting sandbox email, spreadsheet and ticketing services, low-code automation platforms and when to use them.
   2. 08Reliability: idempotency, retries and queues2 hWhy automations run twice, idempotency keys, retries with backoff, dead-letter handling, background jobs and queues, recording each run’s state in PostgreSQL, partial failures and compensation.
   3. 09Tool servers, least privilege and doubt-clearing2 hExposing tools through the Model Context Protocol (MCP), scoping credentials per tool, separating read and write tools, keeping secrets out of prompts; open questions.

   Practice

   Connect the course project to a sandbox ticketing API and a database, trigger it from a webhook, and prove with a test that a repeated event does not create a duplicate ticket.

   Assessment

   Project milestone 1: a webhook-triggered automation with two integrations, idempotent writes and least-privilege credentials, reviewed with feedback.

   You will be able to

   You can connect AI automations to real systems reliably and with only the access they need.

4. 04Human approval, security and privacy+2 live sessions · 4 hours · Keep people in control of consequential actions, and protect the data and systems an automation can reach.

   Learning objectives

   • Classify actions by consequence and decide which require human approval
   • Build draft-then-approve flows with clear previews and an audit trail
   • Defend against prompt injection arriving through emails, documents and tool results
   • Apply consent, data minimisation, redaction and retention rules to automation data

   Live sessions

   1. 10Approval gates and audit trails2 hClassifying actions as read, reversible or consequential; draft-then-approve flows; showing the approver exactly what will happen; timeouts and escalation; recording who approved what and when; letting people correct and override the automation.
   2. 11Security and privacy for agentic systems2 hIndirect prompt injection through emails, web pages and tool results; never letting untrusted content grant new permissions; data minimisation and consent; what is sent to which provider and its retention settings; redacting personal data in logs; India’s DPDP Act 2023 in outline; doubt-clearing.

   Practice

   Add an approval step before the course project sends any reply or changes any record, then attempt guided prompt-injection attacks through sample enquiries and fix what succeeds.

   Assessment

   Assignment: an approval-flow demonstration with its audit log, and a security and privacy self-review with evidence of the defences you added.

   You will be able to

   You can design automations where people approve what matters and untrusted input cannot take control.

5. 05Testing and monitoring automations+3 live sessions · 6 hours · Prove an automation works before release and notice quickly when it stops working afterwards.

   Learning objectives

   • Unit-test tools and test workflow logic with mocked model responses
   • Build scenario evaluations that check outcomes and the steps taken to reach them
   • Trace each run and track usage, cost, latency and failure rates
   • Set alerts, handle incidents and pause an automation safely

   Live sessions

   1. 12Testing agentic systems2 hUnit tests for tools, mocking model responses for deterministic tests, scenario sets with expected outcomes, checking the tools called and their order, testing approval and refusal paths, running evaluations in CI and on every prompt or model change.
   2. 13Monitoring, cost control and incident handling2 hTracing model and tool calls, structured logs without personal data, usage and cost per run, dashboards and alerts on failures and spend, sampling runs for human review, a pause switch, handling and learning from incidents.
   3. 14Testing and monitoring review and doubt-clearing2 hReviewing each learner’s test suite and traces; diagnosing a failing automation from its traces alone; open questions.

   Practice

   Add unit tests, mocked-model tests and a fifteen-scenario evaluation to the course project, add tracing and a cost alert, and diagnose a seeded failure from the traces.

   Assessment

   Project milestone 2: a test suite and evaluation that run green from a clean checkout, with tracing and alerting in place.

   You will be able to

   You can show evidence that an automation works and detect, diagnose and contain it when it does not.

6. 06Business automation project and final assessment+2 live sessions · 4 hours · Apply the whole course to a business automation project, then present and defend it in a live review.

   Learning objectives

   • Turn a written brief into a scoped automation plan with clear risks and approval points
   • Apply the course’s reliability, safety, privacy and testing practices to a new process
   • Present and defend design decisions and known limitations in a live review

   Live sessions

   1. 15Project build lab and mid-point review2 hGuided build time on the final project with instructor support; a short review of each project’s process analysis, approval gates and evaluation plan; agreeing what to fix before submission.
   2. 16Project presentations and final assessment2 hLive project demonstrations with evaluation results and feedback; final assessment across the whole course (pass mark 60%); next steps for continued learning.

   Practice

   Build the final project with regular commits, a working demonstration, evaluation results and an up-to-date README and privacy note.

   Assessment

   Final project submission and live review, and the final assessment (pass mark 60%).

   You will be able to

   You can independently analyse a business process, automate it safely with AI and explain your decisions.

Teaching formats and fees

Every format teaches the full syllabus live, with the same projects, assessments and certificate rules. The fee is per student for the whole course.

Launch fee — GST treatment to be confirmed.

Fees are set in Indian rupees. Amounts in other currencies are estimates; the amount and currency you would pay are shown before payment once online enrolment opens.

Schedules and batches

Choose the schedule that fits your week. Both teach the same syllabus and the same 32 live hours; only the number of weeks differs.

• Weekday Intensive classes are one hour, so each 2 h syllabus session is taught across consecutive weekday classes.
• Weekend Evening classes run on Saturday and Sunday evenings; the exact time and time zone are set before a batch is confirmed.
• Public holidays are not class days unless the instructor confirms a class. Holidays, instructor leave and cancelled classes move the end date later so no live hours are lost.
• Plan about 4 hours a week of exercises, assignments and project work outside class.
• Group, Duo and One-to-One courses all follow one of these two schedules.

Final project

Certificate

Tools and skills

Where these skills apply

• Automating support, sales and operations processes for businesses
• Building internal assistants that work with company systems
• Integrating AI into existing workflow and ticketing tools
• Backend and integration development
• Process analysis and automation consulting for small and medium businesses
• Maintaining, testing and monitoring existing automations

We help you prepare for work; we do not guarantee jobs, placements or salaries. Career resources

Questions about Agentic AI & Workflow Automation