Sridhar Vembu says one smart person using AI can now do the work of 10 people. Most people heard "10". I want you to hear "smart". The headcount math is scary, but the real message is about judgment. AI can multiply a person who knows what good work looks like. It does not yet replace that person.
What Vembu actually said
On 29 October 2026, the Zoho founder posted on X:
"Any job performed on the computer has this risk now: one smart person using AI can perform the work of 10 people. The risk is getting ever more real."
He then added something that got less attention. "This does not mean 90% of people working on a computer will have no work at all." He pointed out that industries that previously could not afford software engineers or digital marketers could afford them now. What it does mean, in his words, is "a rebalancing across the economy and that is painful when we go through it. The only thing we can do is to accept that it is coming and be prepared for it." (Reported by The Indian Express and India Today.)
This came a little over two weeks after he said Zoho would not give its usual pay increase this year. As Mint reported, he cited an uncertain software outlook, a fourfold rise in server prices and rising AI costs. He also wrote, "We will work hard to avoid mass layoffs, but the outlook for software jobs in general is grim. We have not been growing our employee count for 18 months now."
Why I am paying close attention
I used to work at Zoho. So when Vembu talks about work, I listen carefully. I do not have any inside view of Zoho's AI plans today, and I will not pretend to. But I do know he tends to say plainly what many leaders only say in private. That makes his posts worth reading slowly, not just reacting to.
And read slowly, this post is less about "10" and more about "one smart person".
The "10" is the headline. The "smart" is the lesson
Think about what "one smart person using AI" actually needs to do:
- Decide what the real problem is.
- Give the AI the right context.
- Notice when the output is wrong, incomplete or missing something important.
- Take responsibility for the final result.
AI speeds up the middle. The beginning and the end still belong to a person. That person is the multiplier.
The evidence: AI agents still miss what matters
On 27 October 2026, Surge AI published DAYJOB: Tax, a benchmark of 50 realistic tax assignments built and reviewed by tax experts, with 1,789 files like returns, ledgers, receipts and emails. The best model passed only 10.8% of tasks (at a strict 95% rubric threshold). The next two scored 9.6% and 9.2%.
Their examples are the interesting part. In one, a client asked which of two dates to sell her condo. The agent compared both dates correctly but missed a deadline after which she could lose a large tax exclusion. In another, an agent advised a couple on splitting retirement accounts while assuming they were not legally married, even though their own records raised doubts.
Surge's own summary: much of professional work is "figuring out what the problem actually is." The agents could do individual steps. They struggled with judgment.
That is the gap the "one smart person" fills. It is a US tax benchmark, so do not over-read it for every job. But the pattern matches what I see in every team I work with. AI drafts fast. Humans catch what matters.
If you are worried, here is what to do
1. Move from doing the task to directing the task. Pick one recurring piece of your work. Do it with AI this week. Write down where you had to correct it. Those corrections are your value.
2. Get good at checking. Keep the source open. Verify numbers, names and dates. Ask, "What did the client not ask that they should have?" That single question is where the DAYJOB agents failed.
3. Learn the business, not just the tool. Prompts change every few months. Understanding customers, money and risk does not.
4. Show your work. Build a small portfolio: three real problems, how you used AI, what you caught, and the result. This is proof no title can give you.
5. Look where Vembu is pointing. Small businesses, local industries and non-tech firms that could never afford a developer or marketer may now hire one person who can do a lot with AI. In India that is a large opportunity, especially outside the big metros.
This connects to the bigger shift
Earlier today I wrote about the AI layoff debate and one-client risk. Vembu's post is the other side of that story. Companies will need fewer people for the same output. The people they keep, and the new firms that hire, will want the ones who can direct AI and own outcomes.
My take
Do not try to be one of the 10. Try to be the one. Not because you type faster prompts, but because you know what good looks like, you check, and you take responsibility. That is still rare, and today's evidence says it is still needed.
If you want a structured way to build that skill, with real projects and feedback, take a look at our programs at The Arthi AI Collective: thearthiaicollective.com/programs.
Arthi Rajendran, The Arthi AI Collective