I have noticed the lack of movement in this area, perhaps for two reasons; the complexity of compliance in a global environment, and the vastly different processes required in all countries. Can this be reconciled? for sure. How? :) very carefully.
My university used to have a nice custom system to register for courses. They moved it to workday and now it’s an unusable ugly POS that makes it very difficult to plan a schedule. Can’t wait for it to be replaced with something better.
You're right about Workday being exposed but wrong about the replacement.
Your RFP describes a better Workday. It doesn't describe what AI does to business software.
ERPs exist because compute couldn't reason about meaning. To get a purchase order to mean the same thing across procurement, finance, and inventory, someone had to encode one canonical model in advance and bend every company to fit it. The HRIS is the same artifact, applied to people. Workday is a cathedral of encoded rules.
AI changes all that. An LLM is a model of the world. It can render a fitted model per transaction, per jurisdiction, per situation, from raw context, in real time. The encoded canonical model stops being a precondition for the data being processable.
So HRIS doesn't get rebuilt. It splits into three things that were previously fused: a model that's rendered on demand, a system of record that gets smaller, more fragmented and dumber, and an execution engine that gets simpler and deterministic.
The big ERPs will likely manage to rebuild themselves for AI but will still miss the real disruption that just-in-time modelling creates.
The bet shouldn't be on a faster Workday. It's on the substrate underneath: the layer that gives the agent clean, structured, jurisdictionally-accurate context to render its model from.
You're funding the application. The asset is one layer down.
The services cartel point is the sharpest observation in the piece and the one Workday would least want highlighted. 10,500 certified consultants across the Big Four and 150+ shops aren't defending Workday because the product is good. They're defending their own billable hours. The switching cost isn't in the software. It's in the army of people whose careers are built on configuring it.
That's also why the AI-native challenger has to solve two displacement problems simultaneously. Convince the CHRO the new system works better. And somehow make the Accenture partner who bills $2 million a year on Workday implementations enthusiastic about recommending the replacement.
The Flex Credits framing is devastating. Both sides booking the same dollar as "AI investment" so the CFO's dashboard and the earnings call both get what they need. That's not adoption. That's accounting choreography performed by two parties who need the same illusion for different audiences.
The "Open" property is the right wedge. From the carrier/TPA side, we're seeing the same pattern: HR teams standing up their own MCPs to pull benefits and absence data out of Workday because the official path through Studio is a 12+ month build. The carrier-side AI-readiness reviews are running in parallel, and both sides are converging on the same question: why is the integration layer the slowest, most expensive part of the stack?
Workday is exactly the kind of incumbent mid-market HR teams are now questioning at renewal time. The question isn’t just ‘does it work’ but ‘can something AI-native do this for less?’ Great framing.
Joe, I wonder if the bold version of this thesis is not to build the next Workday, but to build the trusted HR control plane that makes the HRIS less central.
If agents become the primary users of HCM, the scarce asset is not the UI, or even the database. It is the governed context around the data. Who can act. What policy applies. What payroll consequence follows. What compliance obligation is triggered. Who approved it. What evidence proves it.
Whoever owns that context layer becomes the real system of action for HR. Workday may still hold the record, but it may no longer be where the work is governed.
Thanks for sharing these insights. You’ve given me some thoughts for the category I’m building which is in a different domain.
Unlike those categories you mentioned, ours is a regulatory and insurability requirement where most incumbents are slapping AI onto legacy systems.
The premise of these incumbents that started 10-15 years ago, was flawed from the start. They built “digital workflows” based on office type environments.
In safety critical, high risk industrial environments where employees interact with atoms, not bits, digital workflows CANNOT BY THEMSELVES, prevent workarounds.
As a bp Manager said to me: “going digital made pencil-whipping easier”.
many will attempt to do this... the ones who want 500,000 companies and 2M plus HR people to try it should think about a big brand like HR.com as their goto market name.
add to the requirements decentralised data portability for employees when they change companies
Love this ! Put the individual profile back into the hands of the person who's profile it is . Universal profile.
I have noticed the lack of movement in this area, perhaps for two reasons; the complexity of compliance in a global environment, and the vastly different processes required in all countries. Can this be reconciled? for sure. How? :) very carefully.
My university used to have a nice custom system to register for courses. They moved it to workday and now it’s an unusable ugly POS that makes it very difficult to plan a schedule. Can’t wait for it to be replaced with something better.
You're right about Workday being exposed but wrong about the replacement.
Your RFP describes a better Workday. It doesn't describe what AI does to business software.
ERPs exist because compute couldn't reason about meaning. To get a purchase order to mean the same thing across procurement, finance, and inventory, someone had to encode one canonical model in advance and bend every company to fit it. The HRIS is the same artifact, applied to people. Workday is a cathedral of encoded rules.
AI changes all that. An LLM is a model of the world. It can render a fitted model per transaction, per jurisdiction, per situation, from raw context, in real time. The encoded canonical model stops being a precondition for the data being processable.
So HRIS doesn't get rebuilt. It splits into three things that were previously fused: a model that's rendered on demand, a system of record that gets smaller, more fragmented and dumber, and an execution engine that gets simpler and deterministic.
The big ERPs will likely manage to rebuild themselves for AI but will still miss the real disruption that just-in-time modelling creates.
The bet shouldn't be on a faster Workday. It's on the substrate underneath: the layer that gives the agent clean, structured, jurisdictionally-accurate context to render its model from.
You're funding the application. The asset is one layer down.
I wrote the longer version of this thesis here, with references to Foundation Capital's context graph work and Tunguz's context database framing: https://datascalehr.com/as-a-founder-this-is-what-keeps-me-up-at-night-tldr-its-about-systems-of-records/
The services cartel point is the sharpest observation in the piece and the one Workday would least want highlighted. 10,500 certified consultants across the Big Four and 150+ shops aren't defending Workday because the product is good. They're defending their own billable hours. The switching cost isn't in the software. It's in the army of people whose careers are built on configuring it.
That's also why the AI-native challenger has to solve two displacement problems simultaneously. Convince the CHRO the new system works better. And somehow make the Accenture partner who bills $2 million a year on Workday implementations enthusiastic about recommending the replacement.
The Flex Credits framing is devastating. Both sides booking the same dollar as "AI investment" so the CFO's dashboard and the earnings call both get what they need. That's not adoption. That's accounting choreography performed by two parties who need the same illusion for different audiences.
great point
The "Open" property is the right wedge. From the carrier/TPA side, we're seeing the same pattern: HR teams standing up their own MCPs to pull benefits and absence data out of Workday because the official path through Studio is a 12+ month build. The carrier-side AI-readiness reviews are running in parallel, and both sides are converging on the same question: why is the integration layer the slowest, most expensive part of the stack?
The whole HR domain starting from recruiting has to be rebuilt almost from scratch. A first stab against the hiring doom loop here: https://syncvio.ai
Workday is exactly the kind of incumbent mid-market HR teams are now questioning at renewal time. The question isn’t just ‘does it work’ but ‘can something AI-native do this for less?’ Great framing.
rank of the rest of least loved software would be great to see lol
I wonder why they have such a small revenue multiple?
Joe, I wonder if the bold version of this thesis is not to build the next Workday, but to build the trusted HR control plane that makes the HRIS less central.
If agents become the primary users of HCM, the scarce asset is not the UI, or even the database. It is the governed context around the data. Who can act. What policy applies. What payroll consequence follows. What compliance obligation is triggered. Who approved it. What evidence proves it.
Whoever owns that context layer becomes the real system of action for HR. Workday may still hold the record, but it may no longer be where the work is governed.
Great article!
Thanks for sharing these insights. You’ve given me some thoughts for the category I’m building which is in a different domain.
Unlike those categories you mentioned, ours is a regulatory and insurability requirement where most incumbents are slapping AI onto legacy systems.
The premise of these incumbents that started 10-15 years ago, was flawed from the start. They built “digital workflows” based on office type environments.
In safety critical, high risk industrial environments where employees interact with atoms, not bits, digital workflows CANNOT BY THEMSELVES, prevent workarounds.
As a bp Manager said to me: “going digital made pencil-whipping easier”.
many will attempt to do this... the ones who want 500,000 companies and 2M plus HR people to try it should think about a big brand like HR.com as their goto market name.
What's the latest on the OPM HCM solicitation award?
Spot on! Let's break the mold and build the future of work ⚡️