On a quiet Tuesday in May 2026, Monday.com quietly became a laboratory for the next great enterprise experiment: selling AI thinking by the drop. The company called it "AI credits." Existing customers woke up to find their email a little heavier, their dashboard a little more colorful, and their pricing model a little more confusing. Basic, Standard, and Pro plans now come with 1,000, 2,000, and 3,000 credits respectively. Go over, and each additional credit costs between $0.01 and $0.0125—25% more if you pay monthly instead of annually.
This is the oldest trick in the metering book. I first saw it in cloud computing. Then in blockchain gas. Now it has arrived in the work management space. The market, which had already knocked Monday.com's stock down more than 50% since the start of the year, responded to the news with a 12.6% relief rally. Investors smelled "AI pivot" and bought the story. But as someone who spent years building tools that translate cryptographic whitepapers into plain language for terrified students, I can smell the difference between a real transformation and a rebranding. This one is real, and that makes it more dangerous.
Monday.com is not just adding an AI feature. It is breaking its own covenant with customers.
Let me rewind. For a decade, Monday.com sold a promise: pay a predictable fee per seat, and your team gets a shared canvas for work. It was called a Work OS. If you wanted more seats, you paid more. It was boring. It was reliable. It had all the psychological comfort of a phone bill. Then came the announcement: the company is now an "AI Work Platform." Native AI agents, one-click connectors to Anthropic, OpenAI, and Microsoft, and non-technical team members can configure workflows without writing a line of code. The value proposition shifted from "organize work" to "execute work." That is not a sentence change. That is an architectural soul change.
In crypto, we would call this a hard fork. The old chain—seat-based SaaS—still exists, but the new chain runs on a different consensus mechanism: consumption. AI credits are not a subscription line item. They are a token, a metered unit of value that the client has to guess, budget for, and ration. And here is where the community question becomes everything.
I have spent over a decade watching communities form around shared ledgers, shared risk, and shared reward. The best decentralized networks don't just align incentives; they align understanding. Monday.com's new pricing model demands a level of customer education that almost no SaaS company is prepared to deliver. When a salesperson had to sell seats, the conversation was simple: "How many people need access?" Now the conversation is: "How many AI agent runs will your marketing team need each month, and what is the average complexity of those runs?" That is not a sales call. That is a course in machine learning operations.
Let's get into the technical underbelly, because this is where the real story hides. An AI credit system is not a billing plugin. It is a real-time resource metering infrastructure. Every agent action, every model call, every tool invocation, every byte of data processed—all of it must be tracked, attributed to a specific tenant, converted into a monetizable unit, and then exposed in a dashboard that a confused finance team can actually read. Building that is not a side project. It is a lightweight cloud billing platform in its own right. From my audit experience with DeFi protocols, I can tell you that building the oracle is often harder than building the vault. Monday.com cut 620–630 people, about 20% of its workforce, and announced $45–55 million in restructuring charges. The remaining team now has to build this metering engine while maintaining the legacy workflows that keep the existing 225,000 customers alive. That gap between the old product and the new product is where trust goes to die.
And I need to say this directly: a company that lays off 20% of its customer success team at the exact moment it introduces a complex metered pricing model is spraying gas on the riskiest fire it has ever lit. Customers will have questions. They will panic about overruns. They will ask for help designing their first AI agent. And the humans who used to answer those calls are gone. That is not a growth strategy. That is a stress test.
Now, let's talk about the unit economics, because the market's 12.6% bounce was based on a narrative, not a spreadsheet. Traditional SaaS margins run between 75% and 85% because the marginal cost of serving another user is near zero. AI credits are different. Every credit burns real compute on an external model—Anthropic, OpenAI, or Microsoft. If the external model cost eats 40–50% of the credit price, Monday.com's blended gross margin begins to slide toward 60%. And here is the ugly, hidden feedback loop that almost nobody in the investor call noticed: AI gets better over time, which means it will consume fewer credits to produce the same outcome. The more efficient the AI becomes, the less revenue the metered model generates. I call this the AI Efficiency Paradox. In crypto, this is like having a blockchain where every successful transaction makes the next one cheaper—great for users, terrible for validators who rely on fees. If Monday.com prices credits based on consumption, then successful AI optimization punishes its own revenue.
But there is another layer. The company still insists on 19–20% revenue growth for the year. With a restructuring charge, a pricing overhaul, and a sales force that must now learn how to sell meters instead of seats, that guidance is more optimistic than a meme coin roadmap. I've seen this pattern before. In 2017, during the ICO frenzy, projects would promise transformative platforms while quietly laying off the people who understood the old code. The white papers were beautiful. The metrics were terrible. The only thing that saved the honest ones was a community that believed in the long-term mission.
So what is Monday.com's community actually worth? It has 225,000 customers. That is not a user base; that is a distributed network of business processes. Every workflow those customers have built is a small treasure of behavioral data—who approves what, when project statuses change, how tasks are routed. That data, if aggregated and used ethically, could create a genuine data flywheel. The AI agents on Monday.com could learn from thousands of marketing teams, thousands of construction teams, thousands of product teams, and get smarter precisely because the community shares the same platform. This is the same logic behind decentralized protocols. The network effect is not in the UI; it is in the shared understanding of work.
But to get there, Monday.com has to solve a trust problem. Enterprise customers are terrified of sending their proprietary workflows to external models. They worry about data isolation, model training, and prompt injection. They worry that an AI agent with too broad a permission set could accidentally delete a critical board or approve something that was supposed to wait. AI agents are not like humans; they are deterministic only until they are not. And for a platform that prides itself on being the "single source of truth," an agent that makes a bold but wrong decision is a catastrophe. The technical challenge is no longer UX; it is controllability and explainability. How does the product make it clear why an AI agent spent 300 credits on a workflow that a human could have done by hand? How does the product let a compliance officer audit an agent's decision history? These are not feature requests. They are existential requirements.
Now comes the part the market doesn't want to hear. The multi-model connector strategy—one-click access to Anthropic, OpenAI, and Microsoft—looks like optionality. In reality, it is a hostage situation. Monday.com is simultaneously a customer of Microsoft's AI backend and a competitor to Microsoft's own collaboration tools. The same dynamic applies to OpenAI and Anthropic: they could decide tomorrow to build their own enterprise agent orchestration layer, and suddenly Monday.com's plug-and-play agent experience becomes a feature that the upstream vendor can replicate with deeper integration and cheaper pricing. In crypto terms, this is the risk of building an application layer on someone else's L1 without a native token. The platform can tax you, fork you, or simply out-compete you in your own lane. Monday.com's valuation rally assumes that the "AI Work Platform" category will be owned by the middle layer. History says the middle layer gets squeezed first.
But let me offer a contrarian defense. Maybe the exact opposite is true. Maybe the metered pricing model is Monday.com's most crypto-native move yet. In the early days of Ethereum, people complained about gas prices. They said it would kill adoption. Instead, gas created a market for block space, and builders learned to optimize. The same could happen with AI credits. If businesses are forced to think about the cost of an AI action, they will design better automation. They will prune unnecessary model calls. They will choose the cheapest model that produces acceptable output. Seven years ago, I watched a group of university students avoid an obvious OneCoin scam because I forced them to think in terms of utility, not hype. Metered pricing forces a similar discipline.
Here is the deeper insight that most analysts are missing: AI credits will not just change how customers pay; they will change how customers feel about paying. When you pay by seat, you are paying for potential. When you pay by credit, you are paying for a result. The psychological contract is different. A customer who burns 500 credits on an AI agent that publishes a flawless weekly report, updates 40 project statuses, and flags three risks before the Monday meeting will not complain about the $5 overage. They will celebrate it. The moment a credit becomes proven ROI, it is no longer a fee; it is an investment. That is the kind of community-building that survives bear markets.
I keep coming back to a phrase I learned in the deepest part of the 2022 crypto winter: Community is the only chain that cannot be broken. It was true for decentralised finance, and it is true for enterprise SaaS. Monday.com can succeed not because of its AI credits, but because of its capacity to make those credits feel like a shared investment in the future of work. That requires radical transparency. Publish a real-time credit burn rate. Provide a break-glass alert before an agent spends more than a certain threshold. Let a customer define an annual cap and never, ever breach it. Give every AI agent a public log that the customer can export and audit. In other words, treat the customer like a validator on your network, not a liability.
The next 12 months will be a pure test of intent. If Monday.com turns AI credits into another way to extract revenue—hidden overages, confetti animations after every purchase, a giant "buy more" button—the community will wake up and leave. SaaS churn is silent until it is loud. Rivals like Notion, Asana, ClickUp, and Microsoft will not wait. But if Monday.com becomes the most trusted steward of AI work execution, the 20% workforce reduction will be remembered as the moment the company got lean enough to build a new category. The 12.6% bounce will turn out to be the beginning of a real repricing.
I don't know which path they will choose. I do know that technology does not create communities. People create communities. And the only way to get a community of enterprises to trust an invisible unit called a credit is to prove, over and over, that the unit represents their success. Do that, and the network effects are impossible to copy. Fail, and the only thing left after the AI hype fades is a legacy work OS with a broken billing model.
This is the fork in the road. Monday.com has already written the code, cut the headcount, and pre-announced the charge. The hard part is what every startup founder eventually learns: the protocol doesn't matter if the people don't believe. They believed once. They can be won again. But it will not be won with a better dashboard. It will be won with a promise, kept. Community is the only chain that cannot be broken.