ChatGPT for Business in India 2026 — Enterprise Rollout Playbook

ChatGPT, Claude and Gemini are no longer "experiments" inside Indian companies — they are line-item budget items. The question in 2026 is not whether to roll out generative AI, but how to do it without leaking customer data, wasting licences, or training people on yesterday's features. This is the playbook I run with Indian enterprises, MSMEs and startups in my corporate AI workshops.

The 5-step rollout that actually works

1. Map the 10 workflows that eat the most hours. Not the most exciting ones — the most repetitive ones. Email drafting, MIS, JD writing, RFP responses, support replies, vendor follow-ups.
2. Pick one tool per category. One chat (ChatGPT or Claude), one search (Perplexity or Gemini), one automation (n8n, Make or Zapier). Three tools, not thirteen.
3. Write a 1-page AI policy. Green / Yellow / Red data. What can be pasted, what can't. DPDP Act 2023 alignment. Every employee signs.
4. Run role-based training. HR, Finance, Sales, Ops and Engineering each get a 3-hour session with prompts for their work, not a generic webinar.
5. Ship a shared prompt library. Notion or SharePoint. Versioned. Owned by an "AI champion" in each team. Reviewed monthly.

Prompt: weekly MIS summary (Finance)

Act as a senior FP&A analyst.
From the table below (weekly P&L, current vs last week vs budget),
write a 5-bullet exec summary for the CFO:
- 2 bullets on top positive movements
- 2 bullets on top concerns
- 1 forward-looking action
Tone: factual, INR in crores, no hedging.
Data: """{paste}"""

Prompt: candidate shortlist note (HR)

Score each resume against this JD on:
1) Relevant experience (0-5)
2) Skills match (0-5)
3) Trajectory (0-5)
4) Red flags
Return a markdown table + a 2-line recommendation per candidate.
JD: """{jd}"""
Resumes: """{resumes}"""

Prompt: customer follow-up (Sales)

Write a 90-word follow-up email to {name} at {company}.
Context: we had a 30-min call on {date} about {pain_point}.
Goal: book a 20-min product demo next week.
Tone: warm, specific, no jargon, no "just checking in".
Include one line referencing {their_recent_linkedin_post or news}.

What goes wrong (and how to avoid it)

• Buying enterprise licences before training. 70% of seats go unused. Train first, scale later.
• No data policy. Staff paste customer PII into free ChatGPT. Use enterprise tiers + DLP.
• One-and-done workshop. Skills decay in 6 weeks without a refresher and a prompt library.
• Ignoring Tier-2 cities. Surat, Indore, Jaipur, Coimbatore teams are hungry for this — don't only train HQ.

Sector snapshots

BFSI: RFP responses, KYC summarisation, internal policy Q&A bots.

Manufacturing: SOP drafting, vendor email automation, shift-handover summaries.

Healthcare: patient-instruction drafting, claims documentation, OPD scheduling assistants.

Education: lesson planning, assessment generation, parent communications.

IT & SaaS: PRD drafts, release notes, support macros, code review assistants.

Want this rolled out in your company?

I run 1-day and 4-week corporate AI programs across India and globally. Email hello@nirmalrabari.inor WhatsApp +91 79902 87281for a custom proposal.