Anthropic CEO Dario Amodei laid out a specific threshold for all of us to expect in his October 2024 essay Machines of Loving Grace: what he calls “powerful AI,” a system smarter than a Nobel Prize winner across most fields and capable of running autonomously for hours, or days, at a time, could be here as soon as 2026 or 2027. OpenAI’s Sam Altman made the same case a year earlier in The Intelligence Age, saying that superintelligence could be “possible in a few thousand days.” Google DeepMind’s Demis Hassabis has moved in the same direction. At Google I/O 2026, he reduced a timeline he’d once put at five to ten years down to “a few years,” and in a separate conversation with Axios described AGI arriving around 2030, plus or minus a year.
Three generative model CEOs, three slightly different dates, with one common trait: every timeline they’ve stood behind has been compressing since 2023. Not everyone agrees though (thankfully), Ilya Sutskever left OpenAI to found Safe Superintelligence Inc. and has refused to name a year at all, and aggregated forecasting communities that track this question tend to predict it to be way past 2030. So, who the hell knows the exact date when the world as we know it will change? It’s coming though, on that they all agree.
What’s Happening Today
None of these men were writing about lead generation. But a system broadly better than humans at almost every cognitive task doesn’t stop at writing and coding. It negotiates. It buys. And the move toward AI agents transacting with other AI agents in B2B commerce is already measurable, well ahead of any superintelligence milestone arriving at all.
Gartner told analysts at its 2025 IT Symposium/Xpo that AI agents will intermediate 90% of B2B purchases by 2028, channeling more than $15 trillion in spend through automated exchanges. So, a buyer’s agent and a seller’s agent engaging, negotiating, contracting, and executing a transaction with minimal, or no, human involvement on either side. A separate Gartner survey of 646 B2B buyers this March found 67% now prefer a representative-free buying experience, and 45% had already used AI somewhere in their most recent purchase, numbers reported in detail by GSPANN’s research team.
Forrester’s forecast for this year alone puts 20% of B2B sellers in a position where they’ll have to respond to a proposal negotiation led entirely by an AI agent on the buyer’s side. Responsive’s October 2025 research goes even further on the discovery side, finding that roughly two-thirds of B2B buyers already lean on AI agents and chatbots as much as, or more than, Google when evaluating vendors, a number that climbs to 80% in tech and software specifically, a stat compiled in Ross Sylvester’s Selling to Agents research roundup. Deloitte’s 2026 research on B2B agentic commerce describes how this will likely play out: AI buyer agents negotiating with AI seller agents, with the friction that’s defined B2B commerce for decades disappearing from the process overnight. In theory, anyway.
Outbound Is Already Being Rebuilt Around Agents
Abhishek Singla, who builds AI sales agent systems for B2B companies at Ziel Lab, published deployment data on this exact topic in April 2026. Across the accounts he measured, human SDR outbound was averaging 2% to 4% reply rates, while well-built AI agent systems were seeing between 5% and 9%, largely from personalization at scale and follow-up timing a human rep can’t sustain across hundreds of accounts. Singla frames this as an augmentation instead of a full replacement, and his data shows AI agents still fall short on relationship-heavy, multi-month sales cycles.
Not every org is waiting for the dust to settle…and potentially be left behind a competitor. Separate 2026 research from Salesmotion found 22% of orgs have already fully replaced their human SDR function with AI, with another 45% running hybrid models. Ramp economics are pushing in this same direction. Bridge Group’s 2026 ramp survey, compiled in a report from Digital Applied, puts time-to-first-meeting for an AI SDR seat at 24 days on average, against 142 days for a newly hired human SDR. The same report cites Forrester’s 2027 predictions, where 63% of revenue leaders expect a single agentic system to own sequencing, research, and reply triage by the end of next year, collapsing a stack of six to eight point tools into just one.
None of this is settled yet, thankfully. A 2026 Mirakl commerce survey found 64% of B2B leaders recognize AI will have a very significant impact on digital sales, but only 20% feel prepared for it, and separate research from BCG puts only 18% of B2B companies at an advanced maturity level in AI commerce readiness.
Regulators Are Already Requiring The Human Layer
While the labs argue over dates, regulators have started writing the human-verification requirement into law rather than leaving it to vendor preferences. The EU AI Act’s Article 50 requires disclosure that AI was involved on every sales call starting in August 2026. France now requires opt-in consent for B2C and B2B cold calls as of this August. US state laws including Texas’s TRAIGA and California’s SB 53 took effect earlier this year, and every one of them tightens restrictions on AI-run outbound. A vendor whose entire qualification process runs through an agent with no person confirming a real buyer exists is operating on shaky ground, both for compliance and for quality.
Scale Doesn’t Remove Fraud.
Darwinium surveyed 500 fraud, risk, and security leaders for its 2026 agentic commerce research and found 97% had already seen AI-facilitated attacks increase over the past year, with the average company now facing more than $7.5 million in combined annual exposure. Agentic commerce cuts both ways; the same infrastructure that lets a buyer’s agent transact with a seller’s agent in seconds also lets a bad-actor-agent do the exact same thing, at the same speed, with none of the friction that used to slow a human fraudster down.
As more of outbound gets handed to agents operating at volumes that no human can keep up with, the qualification that confirms a real human buyer gets MUCH more valuable. A verified HQL, a lead a person has reviewed and confirmed, is the one piece of the pipeline an agent can’t replicate.
What This Means For Choosing A Lead Gen Partner
Superintelligence arriving on Amodei’s, Altman’s, or Hassabis’s timeline is still speculative, and reasonable people can, and do, argue about whether any of those dates is accurate. Agent-to-agent commerce reshaping B2B transactions is not speculative as Gartner, Forrester, Deloitte, and Responsive are all describing the inevitable. Regulators are already writing verification requirements into law, and the fraud data from Darwinium shows that risk is scaling right alongside the potential opportunity superintelligence offers.
Vendors who adopt agent-run lead qualification are optimizing for volume in a world where volume is about to be EASY. Mirakl’s 2026 research shows that 64% of B2B leaders recognize AI’s impact on digital sales, but only 20% feel prepared for it. A live person confirming each lead that you purchase is real is THE differentiator moving forward.