专家眼里路很少，初学者眼里路全开着：Marc Benioff 谈 Agentforce 与 25 年创业心态

Marc Benioff explains that his collection of famous domain names began after ten years at Oracle, when he took a break in Hawaii. He had angel-invested and was fascinated by the internet, so he bought names for ideas he thought would become companies. That habit later gave him Salesforce.com and, eventually, appstore.com.

His relationship with Steve Jobs started in 1984, when Benioff was a 19-year-old Apple intern writing example programs for the Macintosh. Years later, stuck at Salesforce, he asked Jobs for help and drove down with colleagues. Jobs's three instructions were to grow ten times in 24 months, sign a huge customer like Avon, and build an "application economy", which Jobs could not define.

Benioff worked out that Jobs meant an app store and registered appstore.com, then built AppExchange, launched around 2005 or 2006. About a year after the iPhone, Jobs unveiled the App Store to an auditorium of stunned Salesforce staff. Afterward Benioff gave Jobs the URL and the trademark, and Jobs shrugged that the App Store would not be very big.

Benioff describes Jobs as generous and says Jobs disliked enterprise software, once asking how he could tell his kids he sold to CIOs. He recalls that Jobs's last email to him said everything had worked out better than they could have imagined.

Lenny asks about Salesforce's famous 2000 launch, when Benioff hired actors as protesters outside a Siebel conference holding "the end of software" signs. Another group of actors posed as a news crew, and an angry Tom Siebel called the police, though Benioff stresses he admires Siebel. That evening Salesforce held its launch event at a San Francisco theatre.

Benioff says the challenge of getting heard is the same today with Agentforce, which he introduced at Dreamforce. Salesforce has hundreds of customers on it and has moved its own help site to it. He cites a 50% reduction in human escalation and 83% of inquiries resolved automatically.

To spread the word he is testing many approaches at once: ads with Matthew McConaughey and Woody Harrelson, the help.salesforce.com showcase, sales training, aggressive positioning against Microsoft Copilot, and 1,000 to 2,000 new account executives focused on Agentforce. He says he does not know which will work most. The approach is to find the winning tactic and then make it the strategy.

Lenny links this to Seth Godin's idea of being remarkable. Benioff adds the example of Chris Rock, who tests jokes in clubs long before a Netflix special. The lesson for founders is to experiment widely and not be arrogant about what will work.

Benioff has meditated for three or four decades and has been writing software since he was 15. He says he deliberately avoids an expert's mind, because the expert has few possibilities while the beginner has all of them. He warns that phrases like "this is the one thing that will definitely work" signal failure ahead.

He lists six priorities from a recent all-hands: get everyone focused on Agentforce, find fuel for it inside the company, add direct distribution capacity, tell customer stories (he mentions Disney), mobilise the millions of Trailblazers to become "agentblazers", and motivate the 135,000 customers who now have the product to switch it on.

Benioff draws on Japan, where he often visits Kyoto's Zen temples, a habit he says Jobs shared. He tells of bringing Neil Young to Ryōan-ji, where the musician mentally wrote an album while meditating. Benioff's point is that leaders need a place and a practice for clearing their minds, and that place may not be the office.

Asked why Salesforce stays on top despite startups attacking its user experience, he agrees it is too complicated. He says he never looks at the stock price and that the journey is the reward. He still sees himself as a startup CEO and says Salesforce is just beginning the digital labor industry.

Asked for a concrete picture of an agent, Benioff points to the Gap scene in Minority Report, where a store recognises Tom Cruise, remembers his purchases and talks to him. An agent in this sense is not a faceless program but something that works with you. It might live in software, a phone, a home robot or a car that knows your preferences.

He grounds this in his own experience at UCSF while recovering from an Achilles rupture. Routine pre- and post-procedure calls, he thinks, could be handled by agents at lower cost. He says burned-out doctors and nurses spend nights working through digital messages, and agents could free them for deeper face-to-face conversations.

His favourite example is a CT scan, after which no one called to remind him to drink water to flush the contrast. An agent would ask whether he drank the water, took his meds, or needs a repeat lab. Lenny mentions a New York Times story where ChatGPT alone diagnosed better than doctors with or without it, and Benioff says doctors' bias may explain such findings.

Benioff says he keeps having existential moments about AI, and that he has been waiting for this shift for decades. Salesforce expects about 2 trillion AI transactions in the week of the recording through its Einstein and Agentforce platforms, which he believes makes it the largest provider of enterprise AI transactions. He sees this as going faster and more intense.

He describes the path in stages. First automate customer touchpoints across sales, service, marketing, analytics and Slack, integrated with MuleSoft. Second, aggregate data in Data Cloud. Third, put an agentic platform on top, and later a fourth layer of robots and drones. He says Einstein, a name from the einstein.com domain he owned, was essential groundwork for Agentforce.

On jobs, Benioff is candid about his own company. Salesforce needs fewer support engineers because of the Agentforce support layer, while it is hiring more account executives. He expects healthcare to create jobs for which there are not enough people, and a "tale of two cities" where a blue-collar small town is affected much later than San Francisco.

Lenny asks about product-minded founders hesitant to embrace sales. Benioff rejects the framing and uses Agentforce as an example. This was meant to be the year of Data Cloud, but agent technology moved faster than expected, helped by acquiring Airkit, a company led by an entrepreneur whose earlier startup RelateIQ Salesforce had bought. Production Agentforce code shipped at the end of October.

His broader point is that a founder cannot just play the clarinet. Product, sales, marketing, accounting, investors, employees and customers all matter, and a leader must conduct the whole symphony. Narrowing yourself, he says, does a disservice to everyone.

For Lenny's Fail Corner, Benioff describes a layoff of 10% of the company about two years earlier, after pandemic over-hiring. It was Salesforce's first layoff at that scale, and he was bashed in the press and on Twitter. He says the company needed both a financial and a technology transformation, and both were painful.

He quotes his friend Michael Bell, who says "there is no linear success", and compares entrepreneurship to a rock-and-roll roller coaster. On agents, he welcomes competitors such as Google, Microsoft, Oracle and SAP launching their own, because a competitive market makes everyone better, and cites the Japanese idea of Kaizen, continuous improvement.

Lenny suggests a founder may feel exhausted having just absorbed AI and now needing to figure out agents. Benioff says that is exactly the wrong mindset. You want to be eager for the next thing, including the next failure and the next success, because technology keeps getting cheaper, easier and more automated.

He recalls starting on a TRS-80 Model I with 4K of RAM and selling his first software, a program about how to juggle, to CLOAD Magazine for $75 when he was a high school student. His parents had little idea what he was doing. He uses the story to illustrate a growth mindset: ask what the next great success and next great failure will be.

He closes by saying he wants to get to the future first and welcome customers there, and he cites Elon Musk as someone who pursues many future technologies at once. Benioff says he has never seen anything like it.