HTE404
AI for Startups
Faculty
Irakli Chkheidze
Founder at Makesense
Course length
Duration
Total hours
Credits
Language
Course type
Fee for single course
Fee for degree students
Skills you’ll learn
Overview
AI for Startups teaches students to use AI to research an opportunity, build a first product, and run repeatable startup operations. The course follows the AI Triangle: Claude Cowork, Claude Code, and AI Agents. Teams of two or three develop a single startup project across three weeks, progressing from a reusable AI workspace to a live MVP prototype and an agent connected to its tools and data.
The classroom reference build is a startup landing page with an enquiry form, a small CRM, and an agent for follow-up and reporting. Students scope a customer problem, verify AI outputs, test a complete user flow, and delegate tasks with human oversight. The course combines live demonstrations, guided building, and peer critique. Students leave with working artefacts, usability evidence, and a plan for testing their assumptions with real customers.
Learning highlights
- Research and workflows with Claude Cowork. Outcome: A startup workspace with two tested Skills, one connector, and one scheduled task.
- Build an MVP with Claude Code. Outcome: A live prototype with an enquiry form, CRM, and tested user flow.
- Startup operations with AI Agents. Outcome: An OpenClaw agent connected to the project, with an approved action and an automated report.
Course outline
15 classes
Choose a startup problem
Compare Chat, Cowork and Code; choose a customer problem, state assumptions and record a task baseline and access rules.
Research customers and competitors
Build Projects, Memory and instructions around the startup. Verify browser research and introduce browser sub-agents.
Create reusable founder workflows
Create two reusable Skills for research or operations; test them on fresh inputs.
Connect and schedule startup work
Add one connector and run a scheduled task with a checked, visible output.
Demonstrate the workspace
Compare results with the baseline, complete a peer handover and select the prototype's main user flow.
Scope the startup MVP
Research the need; write a concise PRD, user flow and acceptance criteria for the prototype.
Build the landing page and form
Use Claude Code, Explore–Plan–Code–Commit, CLAUDE.md and Git to produce a working first flow.
Capture enquiries in a CRM
Save form records to Neon; build a small admin view and validate inputs.
Test and improve the prototype
Run three peer usability tests, fix issues and manage coding sessions, context and usage.
Launch a working preview
Explain local versus live; deploy and demonstrate the complete database-backed user flow.
Set up a startup operations agent
Choose where OpenClaw runs; configure Bootstrap, workspace context, updates and permissions.
Run agent tasks through Telegram
Pair Telegram and create a custom Skill using the startup project's context.
Connect the agent to the startup
Link GitHub and a prepared data connection; read a record, approve a change and verify the result.
Automate reports and test reliability
Schedule a report; test success, data gaps, conflicting inputs, denied actions and tool failure.
Startup demo and next experiments
Present workspace → live prototype → agent, including the automatic report, a failure handover and next validation steps.
Course materials
Media
Prerequisites
No prior programming experience or AI course is required. Students should be comfortable with basic computer tasks and team projects. Bring a laptop with permission to install software and reliable internet access. Required accounts, installations, and tool usage budgets will be arranged before Day 1. The course uses Claude Cowork, Claude Code, Railway, GitHub, Neon, OpenClaw, and Telegram.
A paid Claude subscription is strongly recommended for the full hands-on experience, as Claude Cowork is not available on free accounts (minimum package: $20/month). Students without a paid subscription can still follow the course and participate in live demonstrations, but they will not be able to complete the Cowork exercises using their own accounts.
Methodology
The course comprises 15 three-hour sessions, totalling 45 scheduled hours. Teams of two or three students work on a single start-up project throughout the module. Sessions combine short lectures, live demonstrations, guided workshops, peer testing, and reflection, with a short break.
Starter files, sample data, and preconfigured connections support the development process. Each week concludes with a demonstration of the workspace, prototype, or connected agent. Students maintain individual learning records and explain their contributions. Peer testing informs usability improvements, while customer validation is treated as a separate next step. Core work is completed in class.
Grading
Apply for this course
AI for Startups
by Irakli Chkheidze
Total hours
45 Hours
Dates
Nov 09 - Nov 27, 2026
Fee for single course
€1500
Fee for degree students
€750
How to secure your spot
Complete the form below to kickstart your application
Schedule your Harbour.Space interview
If successful, get ready to join us on campus
FAQ