Here is the thing worth your attention, and it’s a vendor saying it against interest. Forsys, three weeks ago: the eight-stage CLM lifecycle has not changed — request, drafting, negotiation, execution, obligation management, amendments, renewals, reporting. What has changed is what runs on top of it.
That is Invariant Physics™ stated by a CLM vendor without recognizing it. The process is constant across four technology eras; the technology layer keeps being replaced.
Two obituaries and a resurrection notice
In May 2024 I covered Dr. Elouise Epstein’s argument that the spider kills CLM solution providers — that standalone contract lifecycle management was facing an existential threat from integrated platforms absorbing the function. SAP, Oracle, Workday, Microsoft, ServiceNow. Why buy a separate system to integrate when contract management arrives inside the platform you already run?
Two years on, the AI-era version of the same obituary has a different mechanism. If an agent can read a contract, extract the obligations, monitor them continuously and act on them, then the dedicated container is the least valuable part of the stack. The repository was always the weakest justification for the licence, and agents dissolve it.
Now look at what the vendors are actually doing.
Research published in May 2026 covered fourteen CLM vendors and more than a hundred AI agents between them. One major vendor shipped three agentic features in a single month. Another rebuilt its platform around AI. A third keeps delaying its agentic launch while announcing integrations.
That is a category investing heavily, not a category dying.
So which is it?
The question is wrong, and the wrongness is instructive
Both positions are claims about the technology layer. CLM is dead says the container gets absorbed. CLM is reborn says the container gets smarter. Neither one is a claim about why CLM underdelivered in the first place.
I have written before about the difference between perception and perspective. Perception is what the technology promises. Perspective is what the record shows after it arrives. Both obituary and resurrection notice are perception arguments. They are debates about capability.
So here is perspective.
In January 2023, S&P Global Market Intelligence measured technology adoption across six areas of supplier management. The results:
Highest adoption
Lowest adoption
Contract management — 64%
Risk management — 33%
Supplier onboarding/master data — 63%
Relationship management — 14%
CSR and sustainability — 14%
Innovation management — 7%
⭐ Read that twice. Contract management is the most automated function in supplier management. Relationship management is among the least.
The industry did not fail to automate contracts. It automated contracts more thoroughly than anything else it does — and left the relationship the contract governs running on spreadsheets, email, and whatever the account manager remembers.
That is not a technology gap. It is a choice about what was worth instrumenting, made consistently, across an entire category, over two decades.
What the underdelivery actually looked like
The category did underdeliver, and that is not controversial — it is why both narratives have an audience. Repositories became expensive digital filing cabinets. Capabilities went unused. The renewal date nobody could find in 2015 was still hard to find in 2022, three platforms later.
Here is a vendor’s own 2026 analysis, cutting against interest: over 90% of organizations still require human validation of AI recommendations, because legal leaders want to know why an algorithm surfaced a particular risk. And from another: migration, not licensing, is the single biggest reason CLM projects stall.
Neither is a technology limitation. One is a trust and explainability problem. The other is a data and organizational problem that surfaces the moment you try to move what you already have.
When a category underdelivers, it’s dead and it’s reborn with AI are both more comfortable than the diagnosis that the constraint was never in the software.
That sequence has survived paper, ERP, SaaS, analytics and now agentic AI. Every era replaced the layer running on top of it and left the sequence untouched.
That is what an invariant looks like. Not a prediction — an observation with a failure condition attached. If the underlying process changes when the technology changes, the claim is wrong. Across four eras, it has not.
Which raises the question the buyer’s guides do not ask. If the process has been constant for a quarter of a century and the technology has been replaced four times without fixing the outcome, what exactly is being fixed by replacing it a fifth time?
The part neither thesis touches
Here is where I think the real problem sits, and it has nothing to do with whether the container survives.
A contract is a record of what was agreed. It is not a record of how the work is actually done.
Those diverge immediately and they diverge in both directions. Terms nobody enforces. Accommodations nobody wrote down. Standing exceptions granted verbally in year two that everyone now treats as policy. Escalation paths that exist because the formal one does not work.
Call these the shadow workarounds. They are not failures of compliance. They are the accumulated compensation for wherever the formal design does not fit operating reality — and in most relationships, they are the reason the thing still functions at all.
I have a dated illustration of this, and it is not mine.
In 2005 I gave a keynote to about three hundred automotive industry executives, arguing that organizations spend enormous time and money building RFP processes, supplier catalogues and standing offers, only to have buyers bypass them and default to price at the moment of decision. Partway through, a man in the audience interrupted me — a supplier, and he was not being polite about it:
“I don’t think that they spend any time at all analyzing the RFQs once they’ve sent them out. They go directly to the price auction. We get on a phone call and those who cut the price are those who get the business. You can’t stand here and tell me how much time they spend analyzing the quotes. They don’t. They don’t do dick. That’s the problem.”
I told him I agreed with him completely. What those organizations had built, I said, was a process that delivered none of its promise to the suppliers participating in it, because the buyers did not use it. A futile exercise.
Notice what that is. A supplier, in public, telling a room of executives that the documented process was not the operating process. He was not describing non-compliance. He was describing what actually happens when the specified procedure costs the buyer the outcome the buyer is measured on — and every person in that room already knew it, which is why nobody argued with him.
That was twenty-one years ago. The systems have been replaced four times since.
Now apply agentic AI to that. An agent reading the contract reads the stated agreement. It extracts stated obligations, monitors stated deadlines, and acts on stated terms — faster, more consistently, at greater volume than any human team.
None of that touches the workarounds. It processes the stated version more efficiently.
A disclosure
In 2020 I was engaged for a nominal fee to write articles for a CLM provider on how their platform would help clients through the COVID supply chain crisis. I wrote one. It set out four capabilities: real-time dashboards, AI-based contract management that could extract common terms across an entire contract database, sustaining service levels through better access to contract facts, and stronger financial modelling around DSO impact.
I stopped after the first article. The work was being pulled toward a marketing message rather than an independent assessment, and those are different jobs.
That article described what the technology would make possible. Three years later, the S&P numbers describe what organizations actually adopted — contract management at 64%, relationship management at 14%. The capability arrived. The gap it did not address is still there.
The archive carries zero vendor sponsorships. That was separate paid work, it was brief, and it ended over the same principle this post is about.
The same gap shows up one level higher
There is a serious body of work arguing that the relationship should govern the contract rather than the other way around. Relational contracting, ISO 44001, charter-based models that bring buyer and seller together to establish shared mission, joint governance and open-book frameworks before the terms are settled.
I think that work is directionally right, and it is a genuine improvement on adversarial contracting.
⚠ But it inherits the same limitation, for the same reason. A facilitated session — however skilled, however long — captures what participants can articulate in a room. The workarounds are what people do, not what people say they do. Frequently the person doing one would not name it as a workaround. It is just how the job gets done.
So a charter can codify a relationship that was never the one running, and push the actual operating logic further out of sight, because now there is a signed document it contradicts.
⭐ Which is the same finding I documented in a defence maintenance operation in 1998, arriving through a different door. The relationships producing the failure were on no diagram anyone was trying to optimize. Not because they were hidden. Because nobody had gone and looked.
Tim Cummins is closer to it than the vendors are
Four days ago Tim Cummins wrote that some seek control through authority while he believes in control through intelligence — that when specialists operate in silos, the inevitable result is contention and delay.
He is naming the same thing from the governance side. A silo is a boundary across which the operating reality is not visible. You can put an agent on either side of it and both agents will be right about their own half and wrong about the outcome.
So which is it?
Neither.
CLM is not dead and it is not being reborn. It is being re-platformed for the fifth time against an eight-stage process that has not moved — by an industry that automated the stated agreement to 64% and the relationship it governs to 14%, and is now proposing to make the 64% autonomous.
That asymmetry is the finding. Not the vendor race, not the category’s survival.
The useful question is not which platform, or whether standalone CLM makes it to 2030.
It is whether anyone has traced how the work is actually done before automating the version that was written down.
That is what Phase 0™ is for. It is not a stage in the eight. It is what should happen before you buy the thing that automates them.
-30-
This analysis draws on the Procurement Insights archive — an independent record, carrying zero vendor sponsorships, that I have published openly since 2007 and that consolidates documented client work, lectures, and articles reaching back to 1998 — nearly three decades of contemporaneous observation, gathered in one place rather than created there. Every claim in it is held to the Provenance Ledger™: a verify-before-publish discipline that traces each assertion to a primary source and never quietly edits the record once it is posted. That record is the evidence base for two working lenses — Invariant Physics™, the constant that however far the technology advances, the operating logic must be in place first; and Implementation Physics™, its per-engagement application: the discipline of doing the readiness work before the platform, not after. Getting it right, rather than being right.
Truth Is Believing. Accuracy Is Knowing. Outcome Is Proof.™
So Which Is It? Is CLM Dead, or Is It Being “Reborn” in the AI Era?
Posted on August 18, 2026
0
Here is the thing worth your attention, and it’s a vendor saying it against interest. Forsys, three weeks ago: the eight-stage CLM lifecycle has not changed — request, drafting, negotiation, execution, obligation management, amendments, renewals, reporting. What has changed is what runs on top of it.
That is Invariant Physics™ stated by a CLM vendor without recognizing it. The process is constant across four technology eras; the technology layer keeps being replaced.
Two obituaries and a resurrection notice
In May 2024 I covered Dr. Elouise Epstein’s argument that the spider kills CLM solution providers — that standalone contract lifecycle management was facing an existential threat from integrated platforms absorbing the function. SAP, Oracle, Workday, Microsoft, ServiceNow. Why buy a separate system to integrate when contract management arrives inside the platform you already run?
Two years on, the AI-era version of the same obituary has a different mechanism. If an agent can read a contract, extract the obligations, monitor them continuously and act on them, then the dedicated container is the least valuable part of the stack. The repository was always the weakest justification for the licence, and agents dissolve it.
Now look at what the vendors are actually doing.
Research published in May 2026 covered fourteen CLM vendors and more than a hundred AI agents between them. One major vendor shipped three agentic features in a single month. Another rebuilt its platform around AI. A third keeps delaying its agentic launch while announcing integrations.
That is a category investing heavily, not a category dying.
So which is it?
The question is wrong, and the wrongness is instructive
Both positions are claims about the technology layer. CLM is dead says the container gets absorbed. CLM is reborn says the container gets smarter. Neither one is a claim about why CLM underdelivered in the first place.
I have written before about the difference between perception and perspective. Perception is what the technology promises. Perspective is what the record shows after it arrives. Both obituary and resurrection notice are perception arguments. They are debates about capability.
So here is perspective.
In January 2023, S&P Global Market Intelligence measured technology adoption across six areas of supplier management. The results:
⭐ Read that twice. Contract management is the most automated function in supplier management. Relationship management is among the least.
The industry did not fail to automate contracts. It automated contracts more thoroughly than anything else it does — and left the relationship the contract governs running on spreadsheets, email, and whatever the account manager remembers.
That is not a technology gap. It is a choice about what was worth instrumenting, made consistently, across an entire category, over two decades.
What the underdelivery actually looked like
The category did underdeliver, and that is not controversial — it is why both narratives have an audience. Repositories became expensive digital filing cabinets. Capabilities went unused. The renewal date nobody could find in 2015 was still hard to find in 2022, three platforms later.
Here is a vendor’s own 2026 analysis, cutting against interest: over 90% of organizations still require human validation of AI recommendations, because legal leaders want to know why an algorithm surfaced a particular risk. And from another: migration, not licensing, is the single biggest reason CLM projects stall.
Neither is a technology limitation. One is a trust and explainability problem. The other is a data and organizational problem that surfaces the moment you try to move what you already have.
When a category underdelivers, it’s dead and it’s reborn with AI are both more comfortable than the diagnosis that the constraint was never in the software.
What has not changed in eight stages
Request. Drafting. Negotiation. Execution. Obligation management. Amendments. Renewals. Reporting.
That sequence has survived paper, ERP, SaaS, analytics and now agentic AI. Every era replaced the layer running on top of it and left the sequence untouched.
That is what an invariant looks like. Not a prediction — an observation with a failure condition attached. If the underlying process changes when the technology changes, the claim is wrong. Across four eras, it has not.
Which raises the question the buyer’s guides do not ask. If the process has been constant for a quarter of a century and the technology has been replaced four times without fixing the outcome, what exactly is being fixed by replacing it a fifth time?
The part neither thesis touches
Here is where I think the real problem sits, and it has nothing to do with whether the container survives.
A contract is a record of what was agreed. It is not a record of how the work is actually done.
Those diverge immediately and they diverge in both directions. Terms nobody enforces. Accommodations nobody wrote down. Standing exceptions granted verbally in year two that everyone now treats as policy. Escalation paths that exist because the formal one does not work.
Call these the shadow workarounds. They are not failures of compliance. They are the accumulated compensation for wherever the formal design does not fit operating reality — and in most relationships, they are the reason the thing still functions at all.
I have a dated illustration of this, and it is not mine.
In 2005 I gave a keynote to about three hundred automotive industry executives, arguing that organizations spend enormous time and money building RFP processes, supplier catalogues and standing offers, only to have buyers bypass them and default to price at the moment of decision. Partway through, a man in the audience interrupted me — a supplier, and he was not being polite about it:
I told him I agreed with him completely. What those organizations had built, I said, was a process that delivered none of its promise to the suppliers participating in it, because the buyers did not use it. A futile exercise.
Notice what that is. A supplier, in public, telling a room of executives that the documented process was not the operating process. He was not describing non-compliance. He was describing what actually happens when the specified procedure costs the buyer the outcome the buyer is measured on — and every person in that room already knew it, which is why nobody argued with him.
That was twenty-one years ago. The systems have been replaced four times since.
Now apply agentic AI to that. An agent reading the contract reads the stated agreement. It extracts stated obligations, monitors stated deadlines, and acts on stated terms — faster, more consistently, at greater volume than any human team.
None of that touches the workarounds. It processes the stated version more efficiently.
A disclosure
In 2020 I was engaged for a nominal fee to write articles for a CLM provider on how their platform would help clients through the COVID supply chain crisis. I wrote one. It set out four capabilities: real-time dashboards, AI-based contract management that could extract common terms across an entire contract database, sustaining service levels through better access to contract facts, and stronger financial modelling around DSO impact.
I stopped after the first article. The work was being pulled toward a marketing message rather than an independent assessment, and those are different jobs.
That article described what the technology would make possible. Three years later, the S&P numbers describe what organizations actually adopted — contract management at 64%, relationship management at 14%. The capability arrived. The gap it did not address is still there.
The archive carries zero vendor sponsorships. That was separate paid work, it was brief, and it ended over the same principle this post is about.
The same gap shows up one level higher
There is a serious body of work arguing that the relationship should govern the contract rather than the other way around. Relational contracting, ISO 44001, charter-based models that bring buyer and seller together to establish shared mission, joint governance and open-book frameworks before the terms are settled.
I think that work is directionally right, and it is a genuine improvement on adversarial contracting.
⚠ But it inherits the same limitation, for the same reason. A facilitated session — however skilled, however long — captures what participants can articulate in a room. The workarounds are what people do, not what people say they do. Frequently the person doing one would not name it as a workaround. It is just how the job gets done.
So a charter can codify a relationship that was never the one running, and push the actual operating logic further out of sight, because now there is a signed document it contradicts.
⭐ Which is the same finding I documented in a defence maintenance operation in 1998, arriving through a different door. The relationships producing the failure were on no diagram anyone was trying to optimize. Not because they were hidden. Because nobody had gone and looked.
Tim Cummins is closer to it than the vendors are
Four days ago Tim Cummins wrote that some seek control through authority while he believes in control through intelligence — that when specialists operate in silos, the inevitable result is contention and delay.
He is naming the same thing from the governance side. A silo is a boundary across which the operating reality is not visible. You can put an agent on either side of it and both agents will be right about their own half and wrong about the outcome.
So which is it?
Neither.
CLM is not dead and it is not being reborn. It is being re-platformed for the fifth time against an eight-stage process that has not moved — by an industry that automated the stated agreement to 64% and the relationship it governs to 14%, and is now proposing to make the 64% autonomous.
That asymmetry is the finding. Not the vendor race, not the category’s survival.
The useful question is not which platform, or whether standalone CLM makes it to 2030.
It is whether anyone has traced how the work is actually done before automating the version that was written down.
That is what Phase 0™ is for. It is not a stage in the eight. It is what should happen before you buy the thing that automates them.
-30-
This analysis draws on the Procurement Insights archive — an independent record, carrying zero vendor sponsorships, that I have published openly since 2007 and that consolidates documented client work, lectures, and articles reaching back to 1998 — nearly three decades of contemporaneous observation, gathered in one place rather than created there. Every claim in it is held to the Provenance Ledger™: a verify-before-publish discipline that traces each assertion to a primary source and never quietly edits the record once it is posted. That record is the evidence base for two working lenses — Invariant Physics™, the constant that however far the technology advances, the operating logic must be in place first; and Implementation Physics™, its per-engagement application: the discipline of doing the readiness work before the platform, not after. Getting it right, rather than being right.
Truth Is Believing. Accuracy Is Knowing. Outcome Is Proof.™
Share this:
Like this:
Related