Free with a sign-in.
The best 15 min of 69
Bret Taylor: A New Class of Software Winners
Bret Taylor argues that AI winners will depend on strong products and business models, not just access to powerful models.
The Logan Bartlett Show · Logan Bartlett
- Bret Taylor is CEO of Sierra and chairman of the board of OpenAI.
The best minutes
- 5:59Outcome pricing captures value beyond cost savings8 min
- 18:00Incumbents struggle most with the messy middle of a shiftPro pick 8 min
Also worth watching
- 1:07AI agents expand software into labor markets5 min
- 27:33Applied AI companies need value beyond making the tech work2 min
Skip the funding-round opener and the internet-versus-mobile debate about where platform value accrues.
Full analysis · 6 min read
Real Alpha · Nov 8
Bret Taylor says calls fall from $20 to 20 cents with AI
- Taylor says a call costs $20 today and 20 cents with AI.
- He gives the figures without sourcing them as Sierra results, so the episode does not verify them.
- The business test is whether cheaper service creates customer value, not savings alone.
Free with an account
Read the rest of the analysis
6 sections, the 5 other voices it tests Taylor against, and what it means for you. Every episode we cut gets one.
Check your inbox
We sent a link to your address. Click it to finish signing up. Not there in a minute? Look in spam.
or
Signed up before? The same buttons sign you in. The episode, its moments and the transcript stay free.
Agents make software compete for labor budgets
Taylor’s central market claim is that agents can do a job, not just improve a worker’s productivity. That distinction lets software address labor spend, rather than only software budgets.
For example, he points to legal review and antitrust work. An assistant can expand lawyer productivity, while an agent doing review can be valued against legal labor.
That enlarges the addressable market, but does not establish how much value a vendor captures. Taylor’s internet comparison says the category may be obvious, but winners depend on product, packaging, business model, and execution.
On Silicon Valley Girl in October 2026, Daron Acemoglu compares two cost choices. One boosts 100 workers’ productivity by five percent.
The other automates five percent of jobs and halves those costs. His supplied comparison favors broad productivity over narrow job cuts. It addresses cost-cutting, not whether agents can perform legal labor.
It is not a direct rebuttal to Taylor. It asks founders to measure whether agents create capacity or merely remove labor expense.
Our read
Taylor identifies a larger possible market, not guaranteed paid demand. Buyers should compare agent work with labor costs and the value of making workers more productive.
Outcome pricing can align fees with business value
Taylor’s case for outcome pricing depends on measurable business results. A sale or service cost can connect a fee to value more directly than a generic seat.
He compares that shift to advertising moving from impressions toward cost per click and conversion. The analogy supports a billing model, not a claim that every outcome-priced product keeps high margins.
Logan Bartlett raises the counterpressure, namely alternatives that limit price. Taylor answers that competition may compress fees, but outcome measurement still gives vendors a stronger value reference.
Taylor is selling this logic as Sierra’s CEO, and Sierra builds customer-service agents. He provides no Sierra rate card, deal data, win rates, or customer economics in this exchange.
That absence matters because a fee attached to a result does not show the vendor’s share. Nor does it tell whether another supplier can deliver the same outcome more cheaply.
So the episode supports testing outcome fees against customer value and competition, rather than treating them as a premium by default.
Our read
Outcome pricing offers a way to connect fees with results. It does not establish durable margins, customer willingness to pay, or protection from competitors.
Cheap calls could make service an engagement channel
Taylor provides no customer, timeframe, or source for the comparison, and does not identify it as Sierra data.
Our calculation, $20 divided by $0.20, gives a 100-to-one cost ratio. That calculation clarifies the scale, but does not verify the input figures.
His larger argument is about what firms do with lower costs. For instance, he asks readers to picture a telecom company contacting subscribers more often.
It might encourage plan upgrades or try to reduce attrition. That is a shift from call-center savings toward retention and sales.
Taylor says the shift could change the market, but the episode offers no measured uplift in either outcome. The figures do not establish that more contacts create more value.
Jason Ma’s Dyna Robotics example offers a useful proof standard. On Latent Space in October 2026, he said pilots matter less than customers deploying robots across many sites.
That is a deployment test, not evidence that customer-service calls improve retention.
“the phone call costs $20 today and it costs 20 cents with AI.”
Our read
Cheaper interactions create an option to engage more, not proof that customers will stay or spend more. Test incremental revenue or retention against total service cost.
Incumbents risk the revenue dip before payoff
Taylor labels his numbers even figures. He imagines an account paying $1 million annually and reaching $10 million in 10 years under a new model.
But rolling it out immediately would cut that account to $200,000 that year. Our subtraction is $1 million minus $200,000, an $800,000 first-year decline.
That scenario is not a forecast, but it captures a tension in his argument. A company may accept weaker near-term revenue before a new model matures.
Taylor names quarterly investor pressure, cash burn, employee incentives, and customer-by-customer migration as constraints. He says technology can be hard, yet business-model change is harder in his view.
Rodney Zemmel said this on Inside Blackstone in October 2026. Tech-led AI alone had not produced business impact in his experience.
That supports pairing deployment with business ownership, not Taylor’s specific revenue example. For product and finance leaders, model the transition as well as the destination.
A new pricing model can alter sales compensation, reported recurring revenue, and runway.
Our read
The incumbent’s challenge is not simply choosing a technology. It is funding a migration while keeping customers, employees, and investors aligned.
Applied AI must outlast making technology work
Taylor says early product value can be making a difficult technology work. As platforms mature, that capability can become ordinary, like building a website or scalable database.
His warning is specific to companies building on foundation models. Reliability matters now, but the product needs another source of value after agent capabilities improve.
That leaves an unresolved question rather than a recipe. What will customers continue to pay for? Taylor points to the value SaaS companies built above infrastructure once scaling became routine.
Alex Lieberman described enterprise nuance, inertia, and system wrangling on Latent Space in October 2026. Shawn counters in that episode that data, focused interfaces, and workflow integration can still support durable products.
Those views sharpen Taylor’s test. Durable value may lie in work-specific product and integration, not the temporary feat of completing a task.
Our read
Taylor’s durability test is whether customers value the application after core agent capabilities become easier to access. The transcript leaves the answer to each company’s product and execution.
Cheap calls need proof in customer outcomes
Taylor’s second-order case is stronger than the simple labor-saving pitch. He imagines firms using cheap contacts to protect lifetime value, win upgrades, and reduce attrition.
Yet those are predictions about customer behavior, not demonstrated results in the episode. The stated call costs do not show whether added conversations change subscriber decisions.
A vendor could cut service costs while the buyer sees no gain in sales or retention. An outcome contract needs a result definition both sides can audit.
Richard Socher discussed a customer-rating failure mode on The TWIML AI Podcast in November 2026. His hypothetical bots could game a five-out-of-five rating, showing how a chosen metric can reward the wrong behavior.
That warning concerns measurement design, not proof that cheap service creates growth. The practical test is to track resolution quality, repeat contacts, retention, and sales separately.
Then compare those outcomes with the full cost of running the service. Otherwise a lower bill or a better score can hide weak customer results.
Our read
A lower call cost is an input, not the customer outcome. The case strengthens only if additional conversations measurably improve retention or sales at a repeatable cost.
What it means for you
- Cost per resolved case. Measure the full cost for one workflow, including human escalations and repeat contacts. Compare it with current handling cost.
- Define the billable outcome. Write down what counts as a successful resolution and who verifies it. Track quality separately from volume and satisfaction.
- Model the migration year. Build a near-term revenue and runway scenario alongside the long-term case. Include sales compensation and account-by-account rollout assumptions.
- Test expansion after pilots. Ask for adoption across multiple teams or sites, not only a successful pilot. Track whether deployment produces measurable customer value.
What to watch
- Whether Sierra or its customers publish sourced call-cost and outcome figures.
- Whether service agents increase retention or sales, rather than only lowering operating expense.
- Whether incumbents change pricing and incentives without losing customers or near-term revenue.
Send this to the CFO and customer-service leader sizing an AI agent rollout.
Watch the minute where Taylor gives the call-cost comparison.
Written by Real Alpha from the episode's transcript and the other moments it links, then checked against the transcript. Quotes are the speaker's exact words.
Minute by minute
A Pro write-up, open to you here. Try Pro free for 7 days
A Pro write-up. The minute by minute is in Real Alpha Pro; the episode, its moments and the transcript stay free.
Read it with Pro, 7 days free- 0:15AI markets will reward execution
Taylor says Sierra's new financing is fuel for building a durable company, not an endpoint. He compares AI with the early internet: the markets may be obvious, but he says winners will depend on product, packaging and business model, as he argues Google did compared with AltaVista and Inktomi.
- 4:02Agents expand software markets
Taylor says AI agents can expand software markets by doing work, not just improving productivity. He points to Harvey performing legal work such as antitrust review: that puts a larger pool of legal labor in reach than software sold only to make lawyers more productive.
- 5:59Outcome pricing follows business value
Bartlett asks how competition might limit outcome-based pricing. Taylor contrasts hard-to-measure horizontal tools such as Zoom with department software tied more closely to business value, and compares outcomes-based pricing with the shift from impression ads to cost-per-click and conversion pricing.
- 11:11Cheaper service can grow revenue
Taylor says lower AI costs could change customer service beyond reducing expenses. He gives the example of a phone call falling from $20 to 20 cents: a company could use the savings to have more conversations, sell higher-tier plans and reduce customer attrition.
- 13:54AI could create value across the long tail
Bartlett asks whether AI will benefit new companies or incumbents, as the internet and mobile did in different ways. Taylor says the internet created value across a long tail of businesses, but he finds it difficult to assign credit or predict which software incumbents will persist.
- 18:00Incumbents face difficult transitions
Taylor says existing software companies could keep valuable assets such as their underlying records, but workflows may shift to agents. He argues that incumbents often struggle to change business models, citing Siebel Systems' on-premises software and Microsoft's uneven transition to the internet and cloud.
- 22:25Business-model shifts test leadership
Taylor says changing the business model can be harder than building the technology because incentives, accounting, customers and employees all have to move with it. In a hypothetical example, a customer paying $1 million a year might pay $200,000 during a transition, even if the new product could eventually support $10 million.
- 26:25Applied AI needs lasting value
Taylor says early technology platforms create value by making difficult capabilities work, but those capabilities eventually become commonplace. Applied AI companies therefore need to build value beyond working agents, just as SaaS companies created value beyond the databases they ran in the cloud.
Published Sep 12, 2025.
Transcript · 23 min read
The Logan Bartlett Show, hosted by Logan Bartlett, with Bret Taylor.
From the transcript The Logan Bartlett Show publishes with the episode.
Thank you for doing this. Thanks for having me. I'm glad we got it on the calendar. It's a good time to do it. You guys recently announced a new round. Congratulations. Thank you. I appreciate it. It's a nice milestone on the journey. You could test them at the way you guys have built the date and where you're headed over time. Yeah, it's a milestone, but I think it's a way to put it. You know, raising financing is just, you know, adding fuel to get to the place you want to go. And I think for a entrepreneur like me and my co-founder, Clay, it's, you know, we want to create a enduring, durable company. And so it's just that, just a milestone. But it's good, good time to reflect. We're really proud. I think we're the clear leader in the space we operate, which is AI agents for customer experience, customer service. But I always talk about how similar this AI wave is to the original internet wave and altavisto is first and Google is the one who writes the history book. So we need multiple years of sustained, impeccable execution, basically from here.
And that'll be happy. But I'm excited for the milestone. There's the nor Mcdonald joke that like the good guys always won in history. It's like, you get to define the terms of success in that
“agents aren't simply productivity enhancements for people, but actually doing a job.”
Read the whole transcript
I'm excited for that. I think it's going to be a really positive thing for industry. The other great thing is I can dub over you saying Harvey and slice and LaGoura for the totality of that. And so we don't have to give any plugs to, you know, to hearty throughout this. It's a great market. No, it's a great example too, though. I'm curious,
“the closer you are to really valuable business outcome, the more your platform will be valued relative to the value of that business outcome”
businesses, because you have a great one that you're running here today. But there is this analogy that I've sort of flipped back and forth in my mind a little bit, which is internet versus mobile of what is AI as an opportunity set, does the outcomes set look more like the internet in that the value captured by big independent standalone companies is larger than that of existing businesses that have leveraged the technology in some way, shape or form versus mobile. I think we could probably both agree that the biggest beneficiaries of mobile were probably Google with Android, Apple with the App Store, Facebook, probably most of the value that was created was probably in some type of incumbents in some way.
And I guess as you think about where there might be net new opportunities or where vectors of value are going to be created, do you think that do you think the same about the internet? The most of the value accrued to Amazon and Google as opposed to it's actually I just haven't done the math on it. I've not argued with you actually. It's true with the most valuable companies of all time and you know in the top five of the you know SMB 500 and all that. But it's so interesting. I remember the book The Long Tale and all that but it's it's created incredible economy. So if you took up the top 100 software companies in the stock market, the top five are you know the meta Google, Amazon, etc. Apple, Microsoft. But like the next 40 are like SaaS companies, many of which most people have never heard of unless you're sort of in our business. What is the sum of the long tale? And I look at like incredible businesses like Shopify as an example and you know Shopify is a beneficiary of that. But there are a lot of businesses on top of yeah. I also give like Patrick at Stripe Taco increasing the GDP of the internet.
So I think the long the internet in particular is an extremely long tale and I think it's you know really change the economy in dramatic ways and change distribution mechanics and you know you look at certainly meta TikTok wouldn't exist without the internet and you give credit to the internet or do you prescribe it all there. So I'm happy to answer your question but just to say I don't know actually how to think about that because every company is a digital company at this point. It's interesting. It's an interesting point and maybe I need to reframe how I ask the question because I think you might be right that the totality of the value created on top of the platform itself dwarfs that. It may actually I'd be interested for someone smarter than me to do that analysis. I think it does but I don't know that. It's also hard to define. It's hard to define.
Is it bank with 60,000 software engineers a software company or bank? And the HVAC company that can now reach customers in a more meaningful way that grew from one million revenue to 20 million revenue. Who do you give credit to for that stuff? It's interesting but I guess the question I had is there's sort of this I guess the Salesforce and mobile prove to be Salesforce in a lot of ways. It was kind of a continuing at least within that and obviously there's disruption that ended up happening with Uber and taxis or Airbnb and hotels or whatever it is. I guess as you think about those big software players and you mentioned ServiceNow and SAP obviously you know Salesforce in the ecosystem as well. Do you think that the lion's share of those will be the ones that persist if we picked I don't know the 10 industries and who the whores are not horizontal I guess but like who the application players on there do you think the lion's share of those will survive and be the next generation of it a kind of mobile or do you think we'll look at a lot of disruption and that new companies are going to be who captures ITSM or CRM or ERP or whatever three-letter acronym. I think there will be disruption but I don't think any one of those will definitely
“everyone in the world cares about the messy middle so it requires just incredibly exceptional leadership to go through those transitions”
told this story before but at Google our first campus was sgi silicon graphics and we moved into their campus after they effectively were going out of business and at Facebook we moved into Sun Microsystems campus after they were acquired and sort of like summarily shut down by Oracle and like both of those companies in my relatively short career were successful companies and and build campuses and in my relatively short career at the time had gone out of business and were you know selling their campus for parts you know it was just because their technology and business model was once great and they didn't transition to the next one and that is the default for companies in our industry and there's just a very small handful of companies that have been able to move to multiple products change business models but you can list them on like maybe two hands and I think it's very important as entrepreneurs we keep that in mind because if your ambition is to create a company that outlives you which is certainly my ambition you have to create a culture that's not like a one one trick pony I've heard you talk about like the three different segments of maybe the
AI ecosystem today being financial foundation models tools and then applied AI one of the questions that I think we have come up and I'm not just asking diligence questions that are someone an unanswerable but I think you said that a unique lens is where the foundation models pushing up into different applications where that's going to be true and where it might be harder to go do and you look at something like log code which is kind of saddened between being the foundation model and the actual application itself I guess where you said you have an interesting purview of both you have a board seat with OpenAI and then you also have a company is there any way that you you think about that and like where companies can exist as applied AI systems on top of foundation models versus where foundation models are going to capture value I'll give sort of one framework that I think
“if that continues to be your value proposition from 1998 through 2025 you're not going to exist as a company because it becomes a commodity”
