Marketing, sales, and support have all adopted agents fast. Each works well on its own, but no one knows what the others have learned.
That's not a model problem; it's a context problem, and it costs you in three ways: lower-quality work built on stale or partial understanding, wasted tokens as the agent re-fetches what another has already retrieved, and lost institutional knowledge that stays trapped inside whichever tool produced it.
We’ve seen this before with customer data. The fix was an independent data layer that sat outside the applications. Context is heading down the same path, only faster.
Why Watch?
In this session, Ajay Khanna, CMO at Yugabyte, will share his own experiences building agents for the company and creating a common context layer across the organization.
This session is for marketing, sales, RevOps, and marketing ops leaders, data and platform architects, and others accountable for what agents cost and whether they can be trusted. You'll leave with a checklist you can apply to your agent and context curation projects.
What You’ll Learn
- Why in-app context stores break at enterprise scale, and what to ask vendors before you commit to one.
- What context silos actually cost you, including the token math of agents that start from scratch on every run.
- Five principles for governing context as an enterprise asset.
- How to escape the context paradox: too little context and the model guesses, too much and quality dilutes while cost climbs.
- How the governed context layer works: serving marketing, sales, service, and engineering agents at once.
- A readiness checklist to audit where your context lives today and sequence the fix.
If you want to play along, get the shared context engine here.
