Video summary

Foundations of Agentic AI + AI Workflows | AICTE | BharatCares | Masterclass 4

Main summary

Key takeaways

Educational

Main Ideas, Concepts, and Lessons

1) Program logistics, expectations, and deadlines

Students are urged to:

  • Invite/encourage all team members (not just a representative) to join.
  • Stay attentive and actively participate through to the end.

This session is part of “Masterclass 4” in the IBM SkillsBuild six-week academic internship on AI automation and intelligent solutions.

Submission and deadline guidance

  • Some students may be unable to upload even after completing work; dashboard queries will be addressed.
  • Referenced deliverable deadline: 7th July
  • Final internship project submission deadline: 20th July
  • Submissions must be correct and complete; incomplete submissions may not be evaluated.
  • If a student completed deliverables/certificates but didn’t submit due to issues, tracking may be based on certificate completion date/time.

2) Dashboard and certificate upload methodology (step-by-step)

A dedicated speaker explains how to navigate and upload successfully in the IBM SkillsBuild platform/dashboard.

A. Dashboard link usage

  • Use only the official link repeatedly shared in WhatsApp.
  • Avoid alternative links, as they can break or affect submissions.

B. Learning plan completion and certificate download

After finishing the learning plan sections:

  1. Open the Actions tab (or three-dots menu on mobile).
  2. Select “View completion certificate.”
  3. Download using one of the options:
    • Print
    • Save as PDF (recommended)
    • Save as image
  4. Upload the certificate to the dashboard.

C. Login and common causes of dashboard issues

Login is done using:

  • The email ID used during registration
  • The college name

Common error checks:

  • Using an alternative email can cause attendance/letters/dashboard issues.
  • If one email doesn’t work, try other plausible accounts (e.g., office vs personal), as suggested by the speaker.
  • College names may appear with variations/casing differences.
  • If the account is mapped to NA or an incorrect college entry:
    • Try similarly named college options.
    • If still incorrect, reach out to request correction.

D. Team roles: team leader vs team member

  • Team leader

    • Can make final submissions (e.g., concept note, final project).

  • Team member

    • Can upload only their specific learning plan/certificate for their portion.
    • Team members must not submit in restricted team-leader upload fields (concept note/final deliverables).

E. Certificate upload process (team member)

Steps:

  1. Log in to the correct account.
  2. Go to the student’s section/account.
  3. Click Upload certificate.
  4. Confirm the on-screen caution that the certificate format/identity is correct.
  5. Upload via:
    • Drag-and-drop, or
    • File picker selection

Validation requirements:

  • The certificate must match:
    • Student name
    • Internship track name

If upload fails due to mismatch:

  • Example: uploading a certificate from a different track (e.g., “Big Data track”) can trigger an error like “internship track name is wrong.”

If everything is correct but upload still fails:

  • Reach out to support for investigation.

F. Team leader submission process (high-level)

Team leaders can upload/paste:

  • Concept note (upload or drive link)
  • Final project (upload or drive link)

Upload types and categories can vary:

  • Presentations, screenshots, and code may require different submission categories (upload vs drive link depending on file type).

3) Project planning and Lean Canvas instruction (technical/entrepreneur framework)

The technical speaker (project + Lean Canvas instructor) explains what students must build and how.

A. What the project is (end-of-internship deliverable concept)

  • Students must design the project to be built by the end of the internship.
  • The project must be tied to an AI agentic solution:
    • It must be an agent, not just a static website or generic generative AI tool.

B. What students will learn in Masterclass 4

  1. What project you need to design
  2. How to create a Lean Canvas (one-page business description)
  3. What to submit and how submissions should look

C. Role of UN Sustainable Development Goals (SDGs)

  • Students select from UN’s 17 SDGs.
  • Students choose one SDG and define:
    • a problem within that SDG area.

D. Why Lean Canvas matters

Lean Canvas is positioned as:

  • A one-page description to keep focus on real problems, customers, and business viability.
  • Not “student-only”—students are framed as entrepreneurs.

It helps with:

  • Validating assumptions
  • Building faster
  • Iterating

E. How to fill a Lean Canvas (ordering logic + sections)

The instructor emphasizes it should not be filled purely linearly.

Suggested flow:

  1. Problem
  2. Customer segment
  3. Unique Value Proposition
  4. Solution
  5. Unfair Advantage
  6. Revenue Streams
  7. Cost Structure
  8. Key Metrics
  9. Channels

Lean Canvas sections explained (what to write)

  • Problem

    • Up to ~3 top problems (can be grouped).
    • Must be real pain points that hurt customers.

  • Existing alternatives

    • List what people can use today (e.g., YouTube, Coursera, Udemy, IBM SkillsBuild, ChatGPT, Google search, LinkedIn learning, etc.).

  • Customer segments

    • Primary customers (who will use/buy)
    • Early adopters (who would try it first)

  • Unique Value Proposition

    • What is unique/novel vs existing solutions.
    • Example discussed: “agentic AI mentor” with personalized learning, reminders, progress tracking, etc.

  • Solution

    • Summarize what the product does.
    • Keep it concise (the speaker cautioned against clutter repeatedly).

  • Unfair Advantage

    • Why it can’t be easily copied or bought.
    • Example unfair advantages included:
      • autonomous agents updating from trusted sources,
      • personalized adaptive paths,
      • integrated tutor/mentor/career coach/accountability features,
      • multi-channel engagement,
      • alignment with SDGs,
      • network advantage (example: 40,000 students).

  • Channels

    • Split into:
      • Acquisition (where you find customers)
      • Distribution (how you deliver/access the solution: web app, mobile app, WhatsApp, email, LMS integration, etc.)

  • Revenue streams

    • Examples across B2C/B2B/other:
      • free + premium subscriptions,
      • career coaching,
      • college/university licenses,
      • corporate upskilling,
      • CSR partnerships,
      • certification fees,
      • marketplace/sponsored courses.
    • Instructor notes: structure B2C/B2B/additional clearly.

  • Cost structure

    • Examples:
      • LLM/API usage (model API calls),
      • vector database/hosting,
      • web scraping and databases,
      • WhatsApp/email services,
      • development team (AI engineers, full-stack, UI/UX, DevOps),
      • content validation,
      • customer support,
      • marketing and partnerships/sales.

  • Key metrics

    • Examples listed:
      • daily active learners,
      • completion rates,
      • quiz performance,
      • knowledge improvement,
      • learning consistency/streaks,
      • reminder engagement/open rates,
      • content freshness,
      • retention,
      • career readiness,
      • satisfaction,
      • institutional adoption.

F. Lean Canvas iterations

  • No one creates a perfect Lean Canvas on the first attempt.
  • Lean Canvas evolves as you learn, build, and test.

4) Required deliverables and submission format (what to submit)

The session clarifies submission requirements and what counts as the final project.

A. Deliverables (final)

Submit two things:

  1. Lean Canvas
  2. PPT

Additionally, students must create a working model:

  • The PPT must include a working model link.
  • Include supporting evidence in the PPT (e.g., input/output, screenshots, etc.).

Export format requirement:

  • The Lean Canvas must be exported/submitted as PDF (not PPT).

B. PPT requirements (what the deck should include)

The PPT should work like a project pitch, including:

  • Product overview
  • Screenshots
  • Charts/graphs
  • Process followed
  • Designed agent structures/workflows
  • Evidence it works (reference to the working model)

C. The project must be an AI agent (strict requirement)

  • Only an AI agent will be considered valid.
  • A generic website or only generative AI without agentic behavior is not sufficient.

Agentic requirement is framed conceptually as including:

  • perception,
  • reasoning,
  • acting,
  • learning.

D. Examples of agentic problems across SDGs

Examples discussed:

  • SDG 2 (Zero Hunger): food distribution agent alerts NGOs when excess food exists; personalized WhatsApp messages.
  • SDG 3 (Good Health): medicine/health reminder agent; check-ins based on conditions/preferences.
  • Clean water (SDG 6): monitoring agent using sensors; alerts authorities/cleaning parties.
  • SDG 8 (Decent Work): career mismatch/job skill agent; AI career advisor.
  • SDG 11 (Sustainable Cities): traffic optimization agent connected to mapping/navigation.
  • SDG 12 (Responsible Consumption): waste segregation agent.
  • SDG 13 (Climate Action): energy optimization agent with savings recommendations.

5) Q&A guidance highlights (clarifications from the instructor)

Key clarifications:

  • A working model is compulsory.
  • Lean Canvas is a one-page business description, not a full project report.
  • Team considerations:
    • Typically one Lean Canvas per team (not separate canvases per member).
    • Submission behavior still depends on roles:
      • team leader handles final submissions,
      • members upload certificates only.
  • Deployment and links:
    • Instructor indicated deployment methods will be taught later.
    • Working model can be deployed and shared via a link (web app/host/platform approach discussed informally).
  • Only SDGs from the UN 17 are allowed (with some flexibility mentioned for clubbing SDGs).
  • RAG vs agent:
    • RAG is described as primarily retrieval.
    • Agentic AI implies more action/behavior beyond retrieval.

Speakers / Sources Featured (Identified)

  1. Nandini — organizer/host; thanked speakers; introduces next session; coordinates Lean Canvas sharing and reminders
  2. Ahmed — dashboard navigation explainer; leads certificate upload/dashboard segment
  3. Amit Tari — AI consultant/trainer; leads Masterclass 4 technical session on project design and Lean Canvas
  4. CSRbox team — responsible for dashboard submission guidance and dashboard-related question handling

Additional referenced sources:

  • IBM SkillsBuild (platform referenced)
  • IBM (internship naming and platform context)
  • AICTE (mentioned in internship collaboration)
  • BharatCares CSR box group (mentioned as collaborating organizer)
  • United Nations / UN SDGs (source of the 17 Sustainable Development Goals framework)

Original video