Posted by Why now – A trillion dollar blog
When Foundation Capital article – AI’s trillion-dollar opportunity: Context graphs & Dharmesh Shah wrote about its opportunity, it reinforced adjacent angle that we have been building & executing on. Context Graphs & AI Reasoning make interesting area where big inflexion points will happen.
But context graphs are already red ocean. Yet not all is lost. Larger field of Cognitive Automation has 10s of other fields than context graphs. Some of them we work upon which aren’t context graphs.
Big unlocks – Human Decisioning & Reasoning with AI
Advance AI fields everyone must know
Across AI Infrastructure, there’s a rush to mimic Human Decisioning & Reasoning. That’s the biggest unlock. Many a times called Cognitive Automation.
All the fields under that are adjacant tech or subs stream. Eg Context Graph is just one of them.
Eg. Decision Intelligence System, Autonomous Agents & Agency; Joint Embedding; World Models, Context Graphs; Thinking & more.
Among those, Vaayu works at Decsion Intelligence Layer under Cognitive Automation the most closely. (See a simplistic example of our work below)
If you prefer more techncial style & depth – pls request here.
Context Graph – not the only big thing & way
Context graphs have limitations & they are very hard to acquire plus some new technologies may compete with them. So it’s worth looking at adjacant technolgoies. Eg. Stream that we have been working on.
Eg. Context graph can tell you why a particular decision was taken; But it can’t tell you what should I change now. nor what will be the impact on revenue if such decisions continue to be taken.
AI Reasoning based on World Modelling & Latent State
World & decisions are not linear. This is why most theories fail. The end outcomes are based on various sub systems working together.
Eg. Consider a GTM strategy of a company -> Impacts Top of funnel Traffic -> Impacts Operations -> Impacts load on Customer Support -> Service Ticket Loads -> Impacts brand value, Revenue, Payment recovery
So GTM strategy doesn’t work in isolation. It has latent impact.
What if we could automate above Decision loop with AI
Humans – C-suite reason as the world changes & take action.
Limitations of Context Graph – Example with GTM solved with Vaayu’s Decision Intelligence & AI reasoning
Let’s take an example of a company which is very aggresively preparing for IPO. It wants to own best in world tech, wants to improve metrics, launch to new markets, new segments.
Company has so far deployed traditional Dealer network, offline marketing like trade shows.
AI Reasoning for GTM – Context graphs can’t answer
Decision Intelligence -> Impact on Revenue
So management question is if we used AI – how will revenue improve. They chalk out a strategy to go AI & CRM first. There current channel distributions for AI model training or context are modelled above in pic.
They want to make some changes, budget allocation, team redistribution, introduce AI & heavily use CRM features. This will trigger
So context graphs stop being useful. It can tell you why it happened. But not – what will change.
Context graphs answer:
→ Why did this decision happen?
→ What context led to this outcome?
They do not answer:
→ If we change X, how will revenue change?
→ What should management do next?
From Context Graphs to world models
A world model is an internal, learned representation of an environment that captures:
- State of the system
- Dynamics (how the state evolves over time)
- Effects of actions on future states
GTM Example with Vaayu, after strategy is executed
GTM strategy and execution are not linear.
They are intertwined systems of sales and marketing actions, where each decision influences revenue — often indirectly, and often through second-order effects.
More over, as the GTM unfolds, strategy & planning starts colliding with real world state as of today.
Below shows one of intermittent states of Reasoning Engine when a strategy is rolled out. Green cells show the areas which saw an uplfit. The overall revenue(Green Circle in extreme right)
Decision Intelligence for GTM with AI World models
Eg. Let’s say IPO bound company & management takes 2 actions
- Introduce AI Employees for automation of Pre-Sales workflows
- Rely on digital channels. Redistribute some of marketing budget from offline channels
At micro level – each marketing channel may be called a mini “World Model”. Impact on one channel may have positive or negative impact on other world model.
Eg. AI uplifts prouctivity but say the Average Order value is less on digital channel hence less ROI. But more brand visibility triggers network effect on brand recall side & that leads to more unit sales offline.
This feedback loop & real time actions impact the execution of strategy. AI System learns from live data & past data & predicts future a
Without AI reasoning – this is the work of CMO(Chief marketing officer) & CRO (Chief Revenue Officer). But all these decision happen in their thinking mode trying to understand impact of one thing on another. Many a a times brain gives solution in dreams.
Problems without Decision Intelligence System
→ Deep thinking & Decision mathods are in heads of people.
→ Super costly – Decisions are with single point of failure. Only after execution you may find what happens
Biggest gap of Context Graphs – Latent Impact
World doesn’t work straight & linear. There are second & 3rd order impact. Eg. in above graphs – We haven’t yet covered the skill level of the human sales team. it’s going to impact the end revenue. But we covered how CRM & digital tools lead to productvity gaivs which then had more meetings & demos setup.
This topic we cover in detail in our technical & white papers & future blogs.
Use case examples & Autonmous enterprises-> EV, Quickcommerce, Banking & more
Consider example of EV OEM spending heavily on GTM. This impacts Ops -> Service Tickets -> Payments -> Revenue -> Branding.
Also, we didn’t cover yet how this acts as base of semi to fully autonomus AI employees backed by Reasoning & Decision Intelligence. Which is core theme of Vaayu.
We havn’t covered such examples of latent states & decision loops in this blog. We cover them next separately.
Do you see why we chose name Vaayu – Sales & Finance?
Because of latent impact. Sales impact Revenue. Operations imapct sales & payment collection hence revenue. In abvoe example we showed most simplistic view of GTM Strategy & impact of redistributing the allocation of budget to various channels + adoption of AI Employees.
If you prefer more techncial style & depth – pls request here.
Read more – Vaayu’s past work & jouney towards AI Rasoning & Autonomous systems