I'm proud to announce that @cresta is launching AI Agent Testing 2.0.
A key challenge in AI agent deployments is whether AI agents can be trusted to perform consistently in production, build a testing program rigorous enough to prove it, and maintain that same rigor as the AI agent evolves and scales.
Three places where most testing programs break down are:
- Defining what great looks like in _verifiable_ terms.
- Trusting that your evaluators grade correctly.
- Building coverage that reflects how customers actually behave in practice.
AI Agent Testing 2.0 hands the heavy lifting of testing and evaluation to AI, but keeps human experts in the loop at the most critical points of leverage:
- Requirements are automatically drafted from existing documentation, diagnosed for issues, and refined by domain experts. Those validated requirements become reusable evaluators, calibrated against labeled data so teams can consistently apply expert judgment at scale.
- Test cases generate automatically from production feedback, requirements violations, knowledge base, and the questions customers actually ask.
More than anything else, successfully scaling AI Agents will depend on how confidently organizations can evaluate, improve, and govern them over time.
That confidence is exactly what AI Agent Testing 2.0 delivers: https://t.co/UITYVDfaPs
After @cresta hit $100M in ARR, I sat down with @dougleone and @carl_eschenbach of @sequoia to talk about the impact of customer experience AI on the enterprise, and what’s coming next.
The companies winning with AI aren't automating isolated touchpoints. They're transforming the entire customer experience with a unified context, where every interaction makes the entire system (AI, humans, and the business processes) smarter.
Check out our discussion below.
This is the day I have been looking forward to since I started working on Customer Experience AI almost a decade ago. @cresta has launched Synthetic Customers: realistic, representative customer personas derived directly from an enterprise’s own conversation data.
Every call, chat, email and review contains signals about what customers need, how they behave under pressure, what drives escalation, and what leads to successful outcomes. Synthetic Customers put all of that data to work to help you understand your customers better than ever before.
Synthetic Customers solve multiple problems in the CX AI stack: validate AI agents before deployment via simulation, train human agents with realistic practice conversations via role-play AI Coach Agents, simulate how customer groups may respond to business decisions, and uncover deeper behavioral insights.
Synthetic Customers are not generic personas or artificially constructed profiles. They are grounded in real conversations and organized around representative behavioral patterns observed across the customer base. As customer behavior changes over time, those models evolve alongside it.
Cresta customers can create their own personas by training them from any cluster of customer conversations.
This is a foundational shift in how enterprises understand, test, and prepare for customer interactions. Any company with a continuously updated, evidence-based model of its customer base will make faster decisions, launch with greater confidence, and deliver better customer experiences across every channel.
Learn more: https://t.co/obe6Zeg2pm
Or go to the Presidio, jump in the ocean, get a coffee at The Mill, watch sunset at Twin Peaks, ride a bike anywhere, see live music, eat a burrito, take a grass nap in GG Park, have beer at The Page, watch the Bay Bridge lights, wander Chinatown, wander Ferry building, run across GG Bridge, walk Fort Funston, eat the best meal of your life with friends…drive any direction for 2hrs. And be deeply grateful for the heavenscape you live in.
@btaylor Congrats on the milestone.
I want to get the facts straight. Cresta is the first. We got Level 1 PCI compliant Jan 16, 2026 (see below).
Full audit report is available here: https://t.co/qASnCRCPNw
Respectfully,
Ping
It’s time to return to the place where I know I can have the most impact. I am beyond excited to be rejoining @sequoia as a Partner. Here is what I shared with @gradypb@alfred_lin on how I am approaching my next chapter.
Last Uncapped episode of the year, feeling really grateful for what a fun project this has turned out to be.
Closing out the year with one of my best friends @saammotamedi. Although I don't like to admit it to him, he's a remarkable investor and person.
Greylock has been producing amazing returns for sixty years and I tried to stay serious for long enough stretches of time during this episode to hear some of Saam's insights about how it works. Hope you enjoy.
(0:00) Intro
(1:32) Greylock turning 60 this year
(4:11) What’s persisted since 1965
(8:59) Apprenticeship
(11:34) What's durable in venture
(16:29) Greylock’s ethos
(19:33) Incentive misalignments
(24:44) Breadth vs depth in venture
(29:28) Managing the team on inputs
(34:00) Why incubations are so hard
(43:22) Finding alpha
(52:38) Greylock’s approach to portfolio services
(59:18) Assessing wild revenue ramps
(1:08:10) Horizontal vs vertical SaaS
(1:11:34) Friendships and work
(1:16:26) Saam's biological age 😂
if they ever release old google TGIF’s of larry & sergey, the world would know how fucking hilarious these two dudes were up there every damn friday.
it was like watching a live comedy show all while knocking back few beers at charlie’s.
what a time that was. i don’t think many ppl grasped how unique it was to be a google employee back then.
We’re leading a $300M founding round for @periodiclabs.
Science today is bottlenecked by slow, manual experimentation. Periodic is changing that. They’re building AI scientists paired with autonomous laboratories that can hypothesize, experiment, and iterate at speeds impossible for human-led labs.
Their first target: superconductors and semiconductors that could unlock breakthroughs in transportation, energy grids, and next-gen chips.
This is what happens when frontier AI meets the scientific method. Excited to partner with @LiamFedus@ekindogus and the team as they reimagine how science gets done.
One of the most effective things the U.S. or any other nation can do to ensure its competitiveness in AI is to welcome high-skilled immigration and international students who have the potential to become high-skilled. For centuries, the U.S. has welcomed immigrants, and this helped make it a worldwide leader in technology. Letting immigrants and native-born Americans collaborate makes everyone better off. Reversing this stance would have a huge negative impact on U.S. technology development.
I was born in the UK and came to the U.S. on an F-1 student visa as a relatively unskilled and clueless teenager to attend college. Fortunately I gained skills and became less clueless over time. After completing my graduate studies, I started working at Stanford under the OPT (Optional Practical Training) program, and later an H-1B visa, and ended up staying here. Many other immigrants have followed similar paths to contribute to the U.S.
I am very concerned that making visas harder to obtain for students and high-skilled workers, such as the pause in new visa interviews that started last month and a newly chaotic process of visa cancellations, will hurt our ability to attract great students and workers. In addition, many international students without substantial means count on being able to work under OPT to pay off the high cost of a U.S. college degree. Gutting the OPT program, as has been proposed, would both hurt many international students’ ability to study here and deprive U.S. businesses of great talent. (This won’t stop students from wealthy families. But the U.S. should try to attract the best talent without regard to wealth.)
Failure to attract promising students and high-skilled workers would have a huge negative impact on American competitiveness in AI. Indeed, a recent report by the National Security Commission on Artificial Intelligence exhorts the government to “strengthen AI talent through immigration.”
If talented people do not come to the U.S., will they have an equal impact on global AI development just working somewhere else? Unfortunately, the net impact will be negative. The U.S. has a number of tech hubs including Silicon Valley, Seattle, New York, Boston/Cambridge, Los Angeles, Pittsburgh and Austin, and these hubs concentrate talent and foster innovation. (This is why cities, where people can more easily find each other and collaborate, promote innovation.) Making it harder for AI talent to find each other and collaborate will slow down innovation, and it will take time for new hubs to become as advanced.
Nonetheless, other nations are working hard to attract immigrants who can drive innovation — a good move for them! Many have thoughtful programs to attract AI and other talent. There are the UK’s Global Talent Visa, France’s French Tech Visa, Australia’s Global Talent Visa, the UAE’s Golden Visa, Taiwan’s Employment Gold Card, China’s Thousand Talents Plan, and many more. The U.S. is fortunate that many people already want to come here to study and work. Squandering that advantage would be a huge unforced error.
Beyond the matter of national competitiveness, there is the even more important ethical matter of making sure people are treated decently. I have spoken with international students who are terrified that their visas may be canceled arbitrarily. One recently agonized about whether to attend an international conference to present a research paper, because they were worried about being unable to return. In the end, with great sadness, they cancelled their trip. I also spoke with a highly skilled technologist who is in the U.S. on an H-1B visa. Their company shut down, leading them — after over a decade in this country, and with few ties to their nation of origin — scrambling to find alternative employment that would enable them to stay.
These stories, and many far worse, are heartbreaking. While I do what I can to help individuals I know personally, it is tragic that we are creating such an uncertain environment for immigrants, that many people who have extraordinary skills and talents will no longer want to come here.
To every immigrant or migrant in the U.S. who is concerned about the current national environment: I see you and empathize with your worries. As an immigrant myself, I will be fighting to protect everyone’s dignity and right to due process, and to encourage legal immigration, which makes both the U.S. and individuals much better off.
[Full text, with links: https://t.co/6JNJz88Qyq ]
Giannis has multiple all-time great press-conference answers on his resume. Add this to the list after he was asked about Tyrese Haliburton's father taunting him on the court.
It is not just about the models.
When talking about AI products, the spotlight tends to focus on latest AI model performance against benchmarks. As a subscriber to every major AI assistant on the market, I have been thinking about what really drives the long-term advantage in this space. Here are a few thoughts:
■ Scale Economics
By aggregating potentially billions of subscribers, leaders in consumer ai assistant will have unit economics others cannot match.
◆ Model: amortization of model cost (training and r&d) over a much larger subscriber base.
◆ Compute: more subscribers means better compute resource utilization. Different ways to optimize through the entire stack from batching & caching, to customized chips and owning and operating entire data centers.
◆ Data: favorable unit economics to acquire proprietary IP and data. Data flywheel from more subscribers, especially valuable for long tail subjects and intent.
■ Switching cost
◆ Memory: every token (words, photos, files, etc) into the ai assistant helps build the memory that will make future performance more personal.
◆ Accounts (payment, log-in): as the assistant starts doing things (like Manus), AI assistants will control the browser, sign into websites with user’s credentials and make payments in the future.That’s real switching friction.
■ Network effects:
It is not clear yet how specific network effects are for AI assistants.
In some sense, classic search engine business is a two sided marketplace with traffic and advertisers. For AI assistants, my bet will be from connectors/plug-ins for it to link to commerce directly.
■ “Death star in striking distance”
John Malone referred to Amazon Prime as “the death star in striking distance to every direct-to-consumer company on the planet”. Similarly, I think the winning consumer AI assistant will suck in more and more consumer AI features into its input box.
■ Shipping the org chart → One click is too far
Paradigm shift like AI creates windows of opportunity for startups to get distribution before incumbents get innovation.
However, as shown again and again in history, organizational complexity will inevitably impact the user experience often in the negative way, because, at the end of the day, users do not write performance reviews.