Monday.com CEO on Is SaaS Dead: Will Everything Be Vibe Coded | Eran Zinman

Summary

• The market is pricing Monday.com as if the pivot fails: at the post-earnings low of $70, market cap was ~$3.7B against ~$1.5B cash, no debt, and over $1.3B ARR — roughly a 1.5x EV/ARR multiple. Eran Zinman’s reaction was relief, not despair: “What the market is saying to me is the company is worth zero. Okay, fine. Now I need to build… Screw it. I’m going to go all in.”
• Of the three SaaS doomsday scenarios discussed, Zinman dismisses two and acknowledges the third as a real risk if companies fail to change. Vibe coding is “my least favorite one” (UI ≠ maintainable org-wide software), LLM providers eating the app layer fails the AWS precedent — but on systems of record becoming databases: “It’s a tough one because it’s true… Nobody will want to buy software that not doing the majority of the work for them.”
• His core reframe: software has delivered the same value for 25 years while 90% of the work happened outside the tool; AI flips that to doing 70-80% of the work, so software TAM “is going to be 100x” as spend migrates from headcount (60-70% of a Monday-like company’s expenses) to software (7-8% today). CEOs will “gladly” pay for software that replaces hiring.
• The AI receipts inside Monday are specific: the qualification work formerly handled by ~100 SDRs is now 100% done by agents, with those SDRs moved to outbound (response time 24 hours → 3 minutes, conversion and booking rates all up), support AI-run, every developer on Claude Code and Cursor. Yet Harry’s pushback lands — headcount is still growing mid-teens (down from a planned 20%), and Zinman only commits to roughly flat 3,000 employees by 2030 versus Klarna’s 7,000→2,000 trajectory.
• The pivot is total: biggest transformation since the 2013 launch — product, go-to-market, and pricing moving from seats to hybrid to “eventually 100% consumption.” The bet is horizontal orchestration between humans and agents (“if you haven’t built an agent on Monday, you’re not using the product properly”), plus CRM and service rebuilt from scratch, fully agentic: “Service dominated by ServiceNow, CRM dominated by Salesforce is open market today.”
• Actual AI damage to the business so far is narrow and quantified: Google’s AI mode in search cost ~10% of new-ARR acquisition (high-intent SMB clicks on sponsored links); the other 70 channels saw no impact, and gross retention is the highest ever.
• Skin in the game and capital discipline: an $870M buyback underway, Zinman still holding ~80% of his shares 14 years in, no interest in going private (27% free cash flow, no need to raise). On picking between “OpenAI at 500 or Anthropic at 380,” he reaches for 1998: “Yahoo and Netscape going to suck all the value from the internet revolution — and we know nothing… we’re probably missing 90% of the picture,” though “Anthropic have a little bit more momentum right now.”

Deep dive

1. The sentiment crashed; the business didn’t

• Monday.com sits near $1.3B in revenue yet trades at $3.9B, and Zinman’s opening move is to split the two phenomena: earnings calls look “kind of normal, some companies are even exceeding expectations,” while sentiment shifted “very aggressive, very powerful” in the last six months — “every day there’s a new doomsday scenario, a new tweet.” He’s careful not to dismiss it: “there’s a lot of truth to that.”
• Nor is Monday unique — ServiceNow is down 50%, Salesforce down 60%. His read of what investors actually want: given “infinite demand for AI products,” there shouldn’t be a demand problem, so “you need to prove to us that you can supply that demand… we want to see acceleration of revenue. And they’re right.”

2. Doomsday #1, vibe coding — the one he rates least

• Harry raised one apocalypse scenario; Zinman volunteered the taxonomy: “there’s like five doomsday scenarios I’m familiar with.” Harry joked he could come up with an additional three. The three discussed: everyone vibe-codes their own apps, LLM companies capture enterprise software, platforms decay into systems of record.
• On the CNBC journalist who vibe-coded her own Monday live on air (he found out from a flood of messages as he was going to bed): “she could have done it for Salesforce or ServiceNow, or created a search engine.” His distinction — “a big difference between vibe coding a user interface and building actual software that works across an organization”; people underestimate maintaining software over time, and a dedicated person or team vibe-coding apps can be a huge cost compared with software, which is a small expense for most companies.
• His reductio: if you could vibe-code any company, no startup would have value — yet Ramp just raised at $30B, while Harvey and Legora are also cited. Harry’s pushback: those are trained on proprietary legal data you can’t replicate. Zinman doesn’t concede it: “you can probably get the vast majority of the value even without that specific training.” Verdict: “From all the theories of how software is going to be disrupted, this is my least favorite one.”

3. Doomsday #2 — LLM providers won’t eat the app layer, per the AWS precedent

• The history lesson he leans on: when Amazon built AWS, “everybody said Amazon is going to capture all enterprise software value” because servers were the heavy lifting. “What really happened is the exact opposite” — a boom of companies built on top. LLMs are that infrastructure again, and he expects “an excess of software built on top of the LLM, not less — exponentially more.”
• The mechanics argument: enterprise is “a whole different business” — different sales process, no pure PLG for org-wide deployment, buyers “want to be handheld” — and the labs have a bigger prize chasing the infrastructure layer. “No company on earth, with AI or without, can do everything.”
• Harry agrees this is the weakest doomsday: Anthropic moving into security replacing CrowdStrike or Palo Alto is “a very weak argument” — though he can’t resist “Dario just killed my CrowdStrike over the weekend. Thanks, Dario. Cheers.”

4. Doomsday #3 is the real one: “It’s tough because it’s true”

• No deflection here — “I believe that with all my heart.” His frame: software’s value hasn’t changed in 25 years, from his DOS days onward — a database, dashboards, some workflows. 90% of the work always happened outside the tool: the salesperson makes the call, builds the deck, preps outside the CRM, and only tracks the result inside it.
• AI flips the equation — it “can potentially do 70-80% of the work, and not 10-20% like it used to be.” The consequence, stated flat: “Nobody will want to buy software that not doing the majority of the work for them — otherwise you’re just buying legacy software.”
• His reading of the sell-off is unusually charitable: the market understands the value is changing but “doesn’t know which company will be able to change — and we’re out until we get that confidence. And I get it, because change is extremely hard, especially for an existing company.”

5. The 100x TAM claim — and the headcount paradox Harry wouldn’t drop

• The bull math, from Zinman’s Monday example: on ~$1B of expenses, 60-70% goes to headcount and 7-8% to software. If businesses scale on software instead of hiring, “every CEO will be gladly increasing the expense on software,” and “the TAM of software is going to be 100x from what it is today… the opportunity in the software market today is like we’ve never seen before in our lives.”
• Harry’s pushback — worth keeping: Monday’s own analyst day showed 20% year-on-year headcount growth for ‘26. “Is that not completely paradoxical?” Zinman’s concession: it’s been cut to mid-teens, “we might even take it lower… but I don’t want to press the brakes violently” during the transition.
• Pressed against Klarna’s Seb (7,000 → 3,000 → 2,000 by 2030), Zinman says Monday’s ~3,000 today will “probably not grow substantially” by 2030 — but reframes: slashing headcount won’t change investor perception. “This is the biggest opportunity in software ever in our lifetime… what I want to do is capture as much as I can out of it.”

6. What AI already runs inside Monday

• The qualification work formerly handled by ~100 SDRs is now 100% done by agents, with those SDRs moved to outbound. The stats: callback time went from 24 hours to 3 minutes, missed calls collapsed, “conversion rate went up, people answering went up, opportunity-to-book went up — all parameters went up. AI speaks all languages, available 24/7.” They built the SDR system in-house; support is a mix of third-party and homegrown, because “you still need to highly customize tools today.”
• All developers are on Claude Code and Cursor — the choice between them “right now it’s a sense of flavor.” Output is up, but honestly hedged: “there’s new bottlenecks you find each time you increase one of those productivities” that aren’t attached to writing code.
• The confession that frames the whole pivot: for the first stretch after ChatGPT, “we sprinkled some AI dust on top of our product” — AI formulas, AI blocks, AI columns — “it was sugar coating… the value haven’t changed.” The light-bulb moment came only about a year ago, “not one tweet or one product — a collective understanding that things are changing forever.”

7. The pivot: orchestrating humans and agents, priced on consumption

• “It’s the biggest moment for us in the history of the company” since the 2013 launch — product, go-to-market, homepage, ads, and pricing all change. On pricing his path is explicit: hybrid now, “eventually it’s going to be 100% consumption.”
• The horizontal bet: companies will buy vertical AI tools for SDR, support, legal — but “there’s big room for horizontal plays.” Monday wants to be “the default place for people to build agents and to collaborate between agents and people”: agents output tables, docs, files; humans review and build their own. “If you haven’t built an agent on Monday, you’re not using the product properly… The boards and dashboards are going to be more in the background, the agents more in the forefront.” Scale of ambition: “We’re at 0.001% done of where the world is going.”
• Why ChatGPT, Gemini, and Anthropic don’t win this: they’re personal tools, not thousand-person collaboration surfaces — “it’s like claiming there’s no need for SDR software because you bought Anthropic.” Companies adopting AI “have no idea what they need to do. No idea” — they need guidance on what to automate and what headcount to replace, not an off-the-shelf license.
• The context argument, his sharpest moat claim: “no matter how smart you are, if you don’t have context you cannot perform your job — 90% of the context is not documented anywhere… it kind of floats in the air.” A forward-leaning small team like Jason’s can go full AI; the 95% of the economy that isn’t software companies faces a multi-year transition of humans and agents working together.

8. Two vertical bets into an “open market” — and why incumbents are underrated

• CRM and service are being rebuilt from scratch, 100% agentic. When Monday first entered those markets the competition was brutal; now “the playing field is leveled”: “Service dominated by ServiceNow, CRM dominated by Salesforce is open market today. I believe in their ability to change, but they need to prove it.”
• On whether HubSpot or Salesforce adapts better, an honest non-answer: bigger is harder because “it’s hard to move humans” — and playing offense inverts the usual logic, since SMB customers might adopt new technology before enterprises do. “I don’t know, to be honest.”
• His meta-point on the whole discourse: “every day I go on Twitter… this company has a more robust system of record, this company has a moat — who cares? What are you arguing — who’s going to change last?” The question is offense versus defense, and “I think investors underestimate incumbents’ ability to change” — one public company visibly reaccelerating will flip the cycle.

9. The one measurable AI hit: Google’s AI mode

• The single channel actually damaged by AI: Google’s AI answers cut sponsored-link clicks, accounting for “about 10% of our acquisition in terms of new ARR” — the most transactional, high-intent SMB deals. Budget was shifted to longer-sales-cycle channels; “we have 70 other channels… we didn’t see any impact on other channels.” From his perspective the damage is real but isolated.

10. Skin in the game, the $70 bottom, and cycle discipline

• The morning after the stock hit $70 (having woken at 3 a.m. Monday and 2 a.m. Tuesday): ~$3.7B market cap, ~$1.5B cash, no debt → ~$2B enterprise value on over $1.3B ARR, “like 1.5 multiple.” His reaction: “what the market is saying to me is the company is worth zero. Okay, fine. Now I need to build” — and a feeling of relief. “Screw it. I’m going to go all in… There’s only upside to where we’re at.” Some days, though: “I feel like I was ran over by a truck, hit by a plane, and barbecued — and it’s just 11:00 a.m.”
• Capital allocation: an $870M buyback over the next two-three years, already begun in Q4. Personally he’s constrained by a 10b5-1 plan but has a floor price “much, much lower” than today’s — he’s selling nothing, and still holds “80-something percent” of his shares 14 years in. No take-private: with 27% free cash flow and gross retention at all-time highs, “we don’t need to raise more money… nothing is more sticky than a SaaS product.” M&A is constrained by the valuation inversion — “every startup with 5 million of ARR is now being valued at 2 billion.”
• The discipline came from the top of the last cycle: when the stock hit $400 post-IPO, investor Ken Fox (likely Stripes) told him, “Isn’t it great that the stock is going up without the revenue going up?” — everything gained was sentiment. He didn’t celebrate at 400, and at 70 he says he’ll focus on business metrics rather than the cycle. And being public, he argues, is the advantage over private rivals: “I was kicked in the head so many times by now that — okay, I got the message. When you’re private, you can try and ignore what’s going on.”
• Quick-fire honesty: asked to pick “OpenAI at 500 or Anthropic at 380,” he invokes 1998 — “it’s like saying Yahoo and Netscape are going to suck all the value from the internet revolution, and we know nothing… we’re probably missing 90% of the picture” — while allowing “Anthropic have a little bit more momentum right now.” His changed mind: a year ago “I didn’t understand the magnitude of the change.” The criticism that stings: storytelling — “if people knew what I knew, they would be much more confident in Monday.”