Torani’s $800M AI Test: Can Melanie Dulbecco Keep 500 Jobs?
Most AI “efficiency” plans are just redundancies in a nicer shirt. Torani’s Melanie Dulbecco is betting an $800M business that technology can lift 500 people instead of replacing them.
Most AI “efficiency” plans are just redundancies in a nicer shirt.
Torani CEO Melanie Dulbecco is trying to prove there is another way — and she is doing it with a 101-year-old syrup company expected to generate more than $800 million this year with roughly 500 employees. Her promise is simple, expensive and very easy for other CEOs to sneer at from the comfort of a quarterly earnings call: Torani says it will adopt AI without breaking a no-layoffs record that stretches back to the Great Depression.
That is not a warm-and-fuzzy HR story. It is a management test with real money on the table.
The $800 million promise is not the interesting part
Plenty of companies can grow revenue when the market is going their way. Plenty can also cut heads, call it productivity and get a little clap from analysts for three months.
The difficult bit is building a company that can take a hit without treating the people who created its success as disposable inventory.
Dulbecco joined Torani in 1991, when it was a tiny nine-person business doing about $700,000 in annual revenue. Today, she says the San Leandro, California-based company is on track for more than $800 million in revenue, has averaged 20% annual growth over 35 years, and sells more than 150 flavours across coffee, cocktails and sodas.
The scale matters because this is not a boutique founder telling a lovely story from a beanbag. Torani has factories, supply-chain exposure, café customers, retail customers and the sort of operational complexity that makes keeping people employed during a downturn properly hard.
And yet the company says it has never laid anyone off — not during the Great Depression, not in 2008, not through offshoring and automation, and not when COVID shut down the cafés that buy a big chunk of its product.
Before everyone gets carried away, let’s be adults about this: “no layoffs” is not automatically proof of genius. A business can avoid layoffs by underpaying people, freezing hiring, tolerating poor performance, using contractors as a shock absorber, or simply getting lucky with demand.
But Torani’s model is more serious than that. It combines a long-term employment commitment with employee bonuses tied to both revenue and profit, 401(k) matching, and an employee stock ownership plan for people who have been there more than a year.
That changes the conversation. If staff share in the upside, management has a stronger obligation to explain the downside. Good. That is how it should work.
Melanie Dulbecco’s real decision was made long before AI arrived
The fashionable version of leadership is reacting brilliantly when the crisis lands.
The actual version is making boring, expensive decisions years earlier so you have options when the crisis arrives.
Torani had one of those moments when Starbucks approached it early in Dulbecco’s tenure about becoming a private-label syrup supplier. It was a serious growth opportunity. But the internal debate revealed the trade-off: chasing the work would mean becoming a low-cost producer and sacrificing parts of the business that made Torani distinct, including its use of cane sugar rather than high-fructose corn syrup.
So Torani declined the path.
That is the bit founders regularly butcher. They think every large customer is validation. It is not. Sometimes it is a dressed-up request to abandon the economics, quality or positioning that makes your business worth owning.
A massive customer can turn you into a supplier with no pricing power, no brand and no room to make decisions. You get bigger on paper while becoming weaker in reality.
Torani chose what sort of company it wanted to be. That decision helped preserve the margin, identity and internal trust needed to make the later no-layoffs pledge credible.
The second big decision came during its 2020 move from South San Francisco to a new 330,000-square-foot facility in San Leandro. The company mapped employees’ home ZIP codes and selected a site intended to preserve commutes and retention. Then COVID arrived just as the move was under way.
Most managers love saying people come first. This is what it looks like when it costs something: designing a facility decision around where your people actually live, then moving quickly through a pandemic to keep the business running rather than using chaos as an excuse to clean house.
Torani says e-commerce and retail demand helped fill the gap after café demand fell. Of course they did. But the company could only capture that rebound because it had people, equipment and operational continuity in place.
That is not kindness versus commercial reality. That is kindness used as commercial preparation.
AI is where most CEOs will show their true colours
The phrase “AI will help our people do higher-value work” has become corporate wallpaper. Everybody says it. Half of them mean, “We’ll see how many people we can lose without breaking payroll.”
Dulbecco has put a clearer standard on the table. Her stated question is not merely how Torani can use AI to lower costs. It is what 500 people, or 200 people, or 20 people can do to create more value after technology changes the work.
That is a far better question.
It forces management to do the bit it would rather avoid: define the future roles before installing the technology. Not after. Before.
If AI handles basic reporting, what decisions should the analyst now own? If it removes repetitive customer-service work, how will the team improve retention, sell more intelligently or fix product issues earlier? If it simplifies supply planning, who gets trained to run the exceptions, negotiate better terms or redesign the process?
If your answer is “we’ll work it out,” then you do not have an AI strategy. You have a headcount strategy wearing a robot costume.
Torani calls part of its approach “career mixology”: people moving across roles to build a broader understanding of the business. Fortune cited an employee named Carlos who began in an entry-level supply-chain job, moved through several positions, and now works in procurement.
That model is not quaint. It is exactly the sort of internal mobility companies will need if AI strips out chunks of entry-level work.
The uncomfortable truth is that entry-level roles have always been more than cheap labour. They are how businesses find out who can become exceptional. Remove every basic task without building a new pathway in, and you do not create a leaner organisation. You create a future leadership shortage with better dashboards.
The overlooked angle: job security makes AI more useful
Here is the contrarian bit. The no-layoffs policy is not just an employee benefit. It may be an adoption advantage.
People hide problems from technology programs when they believe honesty will cost them their jobs. They withhold shortcuts, resist process changes, protect their patch and quietly wait for the project to fail. Fair enough, too. You would not eagerly train the machine replacing you.
Torani’s employment record gives staff a reason to participate. If workers believe management is genuinely trying to redeploy them rather than erase them, they are more likely to show where automation can work, where it will fail, and what valuable work can sit on the other side.
That produces better implementation. The people closest to a process usually know where the waste is. They also know where the risk is buried.
I see a version of this in beverage technology while building Agave Finder. The useful technology is not the flashy bit. It is the thing that removes pointless hunting, poor information and repetitive admin so a bartender, retailer, producer or operator can make a better call. If the tech makes knowledgeable people less valuable, you have probably built the wrong thing.
Still, Torani should not get a free pass simply because its values are attractive. The promise gets harder as AI changes white-collar work, consumer demand shifts and the company gets bigger. At $800 million, a bad capital decision or a nasty demand shock cannot be solved by good intentions and a town hall.
The company will need real discipline: cash buffers, sensible hiring, broad retraining, ruthless attention to productivity and a willingness to manage poor performance individually. “No layoffs” must never become “no standards.” That would be unfair to the best people in the building.
But that is precisely why this leadership story matters. Dulbecco is not claiming change will be painless. She is saying management has a duty to make the pain productive rather than simply pushing it downhill onto staff.
What this means for you
If you run a business, do not copy Torani’s slogan. Copy the operating questions behind it.
First, before buying an AI tool, write down the three jobs it will change and the specific higher-value work each affected person will take on. If you cannot name the work, do not pretend this is a growth plan.
Second, build a skills map of your team. Not job titles — actual capabilities. Who understands customers? Who can sell? Who can improve a process? Who can read numbers? Who can run a project? You cannot redeploy people you have never bothered to understand.
Third, make people share in gains they help create. It does not need to be a full employee ownership plan. It can be profit sharing, a properly structured bonus pool, or meaningful incentives tied to outcomes they control. But if all productivity gains go to owners and executives, do not act shocked when your staff treat transformation as hostile.
Fourth, reject growth that wrecks your strategic position. The Starbucks private-label decision is a useful reminder: a big customer is only valuable if it strengthens the business you want to own in five years. Revenue that makes you generic is often a trap.
Finally, tell the truth early. If technology will eliminate tasks, say so. If jobs will evolve, explain how. If you expect people to learn new skills, fund the learning and create the roles. Trust is not built by promising nobody will ever feel discomfort. It is built by proving you will not abandon people the second discomfort arrives.
Most CEOs will use AI to make their payroll line look prettier.
The smarter ones will use it to build an organisation that learns faster, keeps better people and compounds trust. Torani’s $800 million bet is that those things are not soft. They are the whole bloody advantage.