ChatGPT Astra Hindi: Maine AI Ko ₹10,000 Ka Client Dhoondhne Diya

Technological experiment: “Astra AGI” client + automation discovery (₹10,000 target)

• The creator tests GPT-6 Astra (referred to as “GPT Astra / Astra AGI”) by giving it a single goal: Find an AI automation client in Pune where the automation work is worth at least ₹10,000.

• Astra is not told:

  • where to search,
  • which industry/business to choose,
  • what kind of automation to build.
• Astra is also instructed to do no external actions (no emails/forms/purchases), unless the creator explicitly approves.

What Astra does autonomously

1. Independent research using publicly available information.
2. Selects the strongest candidate after researching multiple businesses.
3. Finds a genuine inefficiency/pain point where AI automation adds value.
4. Produces evidence from what the business publicly describes.
5. Recommends an automation workflow, initially designed to handle real-world approval/re-approval constraints.
6. Proposes pricing based on defensibility of the opportunity.

Best-fit business selected (client lead)

• Astra identifies “Property Care” in Pune as the strongest candidate.
• It compares several other businesses (mentions include: Infinite Journeys, Co Thrud Health, Ora Pathology, Bayer and Imperial Overseas, Hinjewadi, etc.) and claims it studied four businesses before choosing Property Care.

Key operational fit:

• Property management involves repairs where owner approval, contractor quotations, inspection findings, and progress updates are part of the workflow.
• Works for owners living elsewhere / NRIs.
• Astra argues the automation aligns with services customers already pay for and likely reduces coordination overhead.

Identified inefficiency → recommended automation concept

Astra’s likely inefficiency

• Repeated work to prepare and track repair approvals and related decision flows, especially when quotations change.

Proposed automation (repair approval automation)

• Turn each repair into a staff-reviewed case record.
• Generate an approval request/brief tied tightly to the exact quotation.
• Track:
  • owner decision,
  • follow-ups,
  • completion updates.
• If the contractor revises pricing, it requires fresh approval.
• It also highlights handling:
  • interpreting inconsistent contractor messages/quotations,
  • summarizing clearly for owner approval.

“Prototype first” requirement (no real integrations / no real data)

• After client discovery, Astra is prompted to:
  • Build and test a working prototype of the proposed workflow
  • using hypothetical/dummy repair cases and simulated inputs
  • without contacting the business or using real credentials/APIs
• The creator’s motivation: avoid touching real customer data while simulating workflow logic.

n8n workflow automation build + browser control demo

• Astra is used in ChatGPT with browser + file operations to generate automation.
• The creator’s system:
  • uses n8n (hosted via Hostinger)
  • Astra attempts to create workflow JSON, import it, and run tests.
• Astra workflow tests include:
  • building workflow logic in n8n,
  • handling file chooser/import interactions,
  • executing a fictional acceptance test.

Acceptance testing: “41 checks across 31 events”

• The prototype runs many validation scenarios, including:
  • normal repair → pass
  • wrong owner → rejected
  • wrong amount → rejected
  • approval required vs revised quote → fresh approval required
  • duplicate events / completion without verification → rejected
• Astra outputs a detailed test report indicating the workflow correctly blocks invalid states.

Output deliverables from Astra (useful for selling)

Astra not only prototypes the workflow but also generates:

• A practical demo plan
• A business outreach draft email
• README / next-steps documentation and file artifacts

Outreach email generated (example content)

• Astra proposes contacting the Property Care team with:
  • an early prototype description using fictional cases,
  • a question about whether it addresses time-consuming coordination,
  • a request for a 15-minute discovery conversation and brief walkthrough,
  • a suggestion of a bounded paid pilot before production.

Pricing suggestion

• Astra initially frames the target as ₹10,000+, but then recommends:
  • ₹18,000 for limited initial implementation
• The creator interprets this as: Astra found enough defensibility to justify a higher price.

Main speakers/sources (as shown in the subtitles)

• Primary speaker: the YouTube creator (Abhijeet / “Abhijeet 99 V2” appears as a coupon code reference in narration)
• System/tool source: GPT-6 Astra / “Astra AGI” within ChatGPT (driving research, workflow generation, prototype testing, and outreach drafting)
• Automation platform mentioned: n8n (browser-hosted instance controlled/imported by Astra)