A second thesis for AI in India
In our previous blogs (read here - 1, 2), we built a thesis on “Orchestration AI”: that was the story of how AI can serve as India's missing project management layer –
We saw how MSMEs and households operate in a "workflow vacuum" - abundant workers and ambitious owners, but no systematic way to coordinate complex operations. Orchestration AI shines here, becoming the conductor coordinating multiple workflows simultaneously (like MyGenie managing construction sites or Kookar orchestrating kitchen operations).
Introducing our 2nd AI thesis
Today, we turn to a different but equally transformative category: Vertical workflow-integrated AI systems that supercharge existing operational efficiency.
Across India's professional services - from small hospitals to educational institutions - there’s a lot of administrative work. Skilled professionals spend 60-70% of their time on these administrative tasks, rather than on their core expertise: doctors spend hours updating treatment notes when they'd rather be with patients, teachers juggle attendance tracking & parent communication instead of focusing on lesson planning.
This thesis sees AI become transformative: not as a replacement for professionals, but as the intelligent layer that handles the stuff they don't want to do. This is what we’re set to uncover today.
Stepping back to understand the AI landscape
Most professional challenges fall into three categories:
Simple tasks are straightforward: spell-check a document, calculate a sum, send a basic notification. These follow clear rules with predictable outcomes. AI utility tools (like AI summarisation or spell check) handle these well.
Complicated tasks: most professional work actually falls into this category. This work involves multiple steps across different systems, but following predictable patterns. E.g.
ICU care requires following protocols & procedures (vitals every 60 minutes, rounds twice a day, etc.);
Teaching a classroom (checking attendance, giving feedback to parents, communicating deadlines).
Each involves moving pieces: different stakeholders, multiple systems, various timelines.
Complex tasks require creativity and judgment: strategic planning, crisis management, and designing a new product. These are unpredictable and emergent - you can't systematise them because each situation is fundamentally different.
Note: This is where AI Assistants (like ChatGPT) can excel. and where Orchestration AI can wrangle order from chaos.
Each of the above require different product approaches:
It’s the middle row of our table - complicated tasks with deep integration - which represents a big opportunity for transformation in India's professional services.
Understanding this framework is crucial because it reveals why most AI solutions miss the mark.
The obvious solution doesn’t work
You'd think AI Assistants like ChatGPT would be the perfect solution. But here's what happens when professionals actually try to use them for workflow coordination:
Even for something simple like designing a personalised fitness program, using ChatGPT becomes an exercise of patience. You need to provide extensive context - what equipment you have, which exercises you actually enjoy, your current fitness level, and how you like your workouts structured. It's painstaking to transfer all this context. And it’s tiring to explain exactly how you want the AI to help.
If a doctor tried using ChatGPT to coordinate patient care, they'd have to design a workflow for ChatGPT to adhere to, and spend their day copying patient information from monitoring equipment to the AI, then editing outputs to fit back into medical record systems. It’s fair to say AI Assistants require you to become the bridge between the AI and your existing workflows.
An integrated fitness AI tool would work completely differently for this use case. Instead of explaining your preferences to ChatGPT each time, it would already be integrated with your fitness ecosystem - your previous workout logs, equipment list, and stated goals would persist in the system.
Why this approach works:
In-workflow context
Unlike AI Assistants that require you to explain the situation every time, Integrated AI systems maintain context within their domains. Granola demonstrates this by preserving meeting context over time. When someone mentions in a meeting, "let's circle back on the budget discussion from last week," you can prompt Granola to pull up what was discussed in Tuesday's planning meeting - it will surface the specific concerns that were raised and what action items were assigned. Rather than manually searching through previous meeting notes, it allows you to query your meeting history and quickly reconstruct the context you need.
The value is in connecting information across related meetings and eliminating the friction of reconstructing previous conversations.
Becoming the coordinator between systems
Take any complicated professional environment - monitoring equipment, communication channels, documentation platforms, scheduling tools, stakeholder management systems. Each operates independently, but the real work happens in the coordination between them.
CloudPhysician demonstrates this coordination beautifully through their AINA system. Their AI video co-pilot analyses patient video feeds along with historical medical records, detecting changes that humans might miss. For e.g., when AINA detects lowered bed rails that indicate fall risk, it automatically sends alerts to the bedside team, coordinating response across monitoring data, patient history, and care protocols.
Integrated AI becomes an invisible conductor that helps them work in harmony. This kind of multi-system integration is simply impossible for AI Assistants operating in isolation.
Enhancement, not replacement
The magic happens when professionals don't have to change how they work - they just get dramatically better results from their existing routines.
CuePilot shows this perfectly in education. Instead of learning new software, teachers can just speak naturally - "Ansh was late, everyone else was present. We're planning a Holi event next Tuesday," - and CuePilot automatically executes multiple administrative actions across fragmented systems, updating attendance, planning events, and communicating with parents.
This is why Integrated AI eliminates adoption friction entirely - professionals don't need to learn dramatically new frameworks, change their workflows, or remember to use the AI. It simply enhances what they're already doing.
This workflow integration approach is fundamentally different from the Orchestration AI we discussed in the previous newsletter. Orchestration AI creates entirely new management capabilities - it steps into the role of project manager. Integrated AI, by contrast, amplifies existing workflows rather than creating new ones. The distinction matters because it changes both the adoption pattern and the value proposition:
Orchestration AI solves the "we need someone to manage this complexity" problem
Integrated AI solves the "we're already doing this work, but it's incredibly inefficient" problem
Both are transformative, but they address different pain points in India's professional landscape. Where businesses lack systematic coordination entirely, Orchestration AI provides the missing management layer. Where professionals are already coordinating complex workflows but drowning in administrative overhead, Integrated AI provides the efficiency layer.
What we're witnessing with Integrated AI represents a fundamental shift in how technology integrates with human expertise. Unlike previous waves of automation that required businesses to adapt to technology, Integrated AI adapts to how professionals already work.
For Indian businesses operating with established processes but fragmented execution, this 1→10 multiplier of Vertical enables dramatic scaling without operational DNA changes. The opportunity is enormous: every predictable but fragmented workflow across Indian businesses becomes a candidate for systematic enhancement.