Most Indian startups approach AI backwards: they buy a tool, try to use it everywhere, get frustrated, and conclude "AI doesn't work for us." The startups winning with AI do the opposite — they start with a specific business problem, choose the smallest possible tool that solves it, measure the outcome, and only then expand. This 90-day roadmap is the framework we walk our clients through. It is concrete, sequenced, and built around what actually works in the Indian startup context.
Month 1 — Audit: Find the High-Value Problems
The goal of Month 1 is not to build anything. It is to identify exactly where AI will create the most value in your specific business. Most startups skip this and waste months building the wrong thing.
Week 1-2: The Operations Audit
Ask every team member (including yourself) to spend one week tracking their daily tasks in a simple log:
Date | Task | Time spent | Repetitive? (Y/N) | Data involved? (Y/N) -----|------|------------|-------------------|-------------------- Mon | Reply to support emails | 2 hrs | Y | Y Mon | Review weekly sales report | 1 hr | Y | Y Mon | Negotiate partnership terms | 3 hrs | N | N
At the end of Week 2, compile all the "Yes/Yes" rows. Those are your automation candidates — tasks that are repetitive AND involve moving or processing data.
Week 3: The Impact-Effort Matrix
For each automation candidate, score it on two dimensions (1-10):
- Impact: If this was fully automated, how much time or cost would it save per week?
- Effort: How hard is it to automate? (1 = simple data routing, 10 = requires custom ML model)
Plot them. The top-right quadrant (high impact, low effort) is where you start. These are typically: lead response automation, report generation, invoice data extraction, social media scheduling, support ticket routing.
Week 4: Choose Your First Project
Pick exactly ONE automation from the high-impact, low-effort quadrant. Write a one-page brief:
Current process: [Step 1 → Step 2 → Step 3 — document the exact manual steps]
Proposed automation: [What would the automated version do?]
Success metric: [How will you know it is working? E.g., 80% of support queries answered without human, report generated in 0 hours vs 3 hours]
Tool to use: [n8n / Make.com / Botpress / Claude API]
Budget: [Monthly tool cost + one-time setup]
Month 2 — Build: Deploy Your First Automation
Month 2 is about building, deploying, and iterating on your first automation. The goal is not perfection — it is a working system that you can observe and improve.
Week 5-6: Build and Test Internally
Based on your project brief, build the automation using the tools you identified. Follow this sequence:
- Build the happy path first — make it work for the standard case (80% of inputs)
- Add error handling — what happens when the input is unexpected or missing?
- Test with 20 real examples — use actual past data, not synthetic test cases
- Document how it works — write a 1-page runbook so any team member can maintain it
Week 7-8: Soft Launch and Monitor
Run the automation in parallel with the manual process for 2 weeks. Compare outputs side by side. Track:
- Accuracy rate: What % of automated outputs are correct without human review?
- Edge cases: What inputs does it fail on? Add these to your improvement backlog.
- Time saved: Track actual hours saved vs the baseline from Month 1 audit.
If accuracy is above 85%, you are ready to make it the primary process. If below 85%, spend another week improving the prompt, the flow logic, or the training data before switching over.
Month 3 — Measure and Scale
If Month 1 is strategy and Month 2 is building, Month 3 is about turning a one-off experiment into a repeatable system.
Week 9-10: Measure Real Business Impact
Go back to your success metric from the project brief. Measure the actual outcome:
| Metric | Before | After | Change |
|---|---|---|---|
| Hours per week on task | Fill in | Fill in | % saved |
| Cost per task | Fill in | Fill in | ₹ saved |
| Speed of output | Fill in | Fill in | % faster |
| Error rate | Fill in | Fill in | % improvement |
Present this data to your co-founders and team. Real numbers create organizational buy-in that no amount of convincing can match.
Week 11-12: Plan Your Next 3 Automations
With one successful automation proven, go back to your impact-effort matrix and select the next 3 projects. Now you have:
- Confidence in the approach (because you have seen it work)
- A repeatable process (your Month 2 build process)
- Data to make the business case (your Month 3 measurement)
Assign one automation per team member or per quarter. Build a rolling 90-day automation roadmap that becomes a permanent part of your operations planning.
Month 2: ☐ Automation built and tested ☐ Parallel run completed ☐ Runbook documented ☐ Deployed to production
Month 3: ☐ Impact measured with real data ☐ Team aligned on outcome ☐ Next 3 automations planned ☐ Rolling roadmap created