“This type of change management was already the hardest discipline in enterprise transformation pre-AI. AI makes it harder…”
Nailed it. AI is increasing the velocity of technical change but the underlying human dynamics are the same (actually - worse, because you have AI fear and fomo thrown in).
It’s an org design problem which is a lot harder to solve.
Thanks for including the Ramp article, they are doing some very interesting things
Lily, I always value your perspective and frameworks. Are you working closely with IT to determine the tech stack? Or perhaps you own that. It seems like the first thing to do is for a company to determine their tools so that AI orchestration can fun cross-department.
That means a lot coming from you, thank you so much for reading!
And yes, this is very much a cross-functional effort. For us, IT owns and gives us access to the core infrastructure (Microsoft Azure, AI Foundry, Claude, Copilot), and we own the MarTech layer on top - Zapier for automation, our own internal tools, etc. A example: my banner generator runs on a Zapier flow using an Azure Function App I built with Claude. Marketers request a banner, it generates 4 ad sizes, drops them in SharePoint, and emails the requestor.
Another piece I'm trying to solve for is data access since it lives everywhere: CRM, Gong, SharePoint, etc. So starting small here as I work with other teams in parallel. First with low risk areas where the data is simple (SEO/AEO, social media) like brand guidelines, internal docs, content the agent just needs to read. You can give AI/agents access to these now without heavy infrastructure. For CRM and other tools, we would partner with IT and systems teams for the tools needed - perhaps Glean, MCP, integrations we need to build. Meanwhile we can run agents and workflows through approved CRM apps and steps within Zapier.
I would say that because the landscape is changing so fast with AI - and how agents access data is also changing quickly (MCP vs. APIs vs. CLIs), so flexibility matters more than picking the "perfect" stack on day one.
I'm super curious what you're seeing on your end? What's been the biggest challenge in your journey to enabling and using AI at an org level?
“This type of change management was already the hardest discipline in enterprise transformation pre-AI. AI makes it harder…”
Nailed it. AI is increasing the velocity of technical change but the underlying human dynamics are the same (actually - worse, because you have AI fear and fomo thrown in).
It’s an org design problem which is a lot harder to solve.
Thanks for including the Ramp article, they are doing some very interesting things
yes! i’ve been following them for a bit. they seem so far ahead of the curve.
Lily, I always value your perspective and frameworks. Are you working closely with IT to determine the tech stack? Or perhaps you own that. It seems like the first thing to do is for a company to determine their tools so that AI orchestration can fun cross-department.
That means a lot coming from you, thank you so much for reading!
And yes, this is very much a cross-functional effort. For us, IT owns and gives us access to the core infrastructure (Microsoft Azure, AI Foundry, Claude, Copilot), and we own the MarTech layer on top - Zapier for automation, our own internal tools, etc. A example: my banner generator runs on a Zapier flow using an Azure Function App I built with Claude. Marketers request a banner, it generates 4 ad sizes, drops them in SharePoint, and emails the requestor.
Another piece I'm trying to solve for is data access since it lives everywhere: CRM, Gong, SharePoint, etc. So starting small here as I work with other teams in parallel. First with low risk areas where the data is simple (SEO/AEO, social media) like brand guidelines, internal docs, content the agent just needs to read. You can give AI/agents access to these now without heavy infrastructure. For CRM and other tools, we would partner with IT and systems teams for the tools needed - perhaps Glean, MCP, integrations we need to build. Meanwhile we can run agents and workflows through approved CRM apps and steps within Zapier.
I would say that because the landscape is changing so fast with AI - and how agents access data is also changing quickly (MCP vs. APIs vs. CLIs), so flexibility matters more than picking the "perfect" stack on day one.
I'm super curious what you're seeing on your end? What's been the biggest challenge in your journey to enabling and using AI at an org level?