Jul 7, 2026 · 1h 1m · a16z
Software in the Age of Agents | The a16z Show
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of The a16z Show, host Elena Burger, a16z partner Seema Amble, and board partner Steven Sinofsky discuss how AI agents and headless architecture are transforming enterprise software. They examine the drivers of software stickiness, the complexity of legacy business logic, and strategic opportunities for AI startups.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →
The host as informed peer 3.6 Guest teaching 4.9 Guest disagreement 1.3 The host pushing back 0.9
speaking balance: gold is the host, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 6:54 Steven's witty reframe of agent terminology
Steven forcefully rejects popular industry hype around agents by cheekily defining an agent as a program that takes too long to run and might not finish, which used to be called a bug.
Hardest push from the host ▶ 14:55 Elena invoking Steven's own essay against market narrativesElena explicitly challenges the popular narrative of software extinction by holding Steven accountable to his own written refusal of the Sasspocalypse.
Biggest teaching moment ▶ 21:40 Steven's Goldman Sachs Excel anecdoteSteven educates the room on the true nature of software stickiness by showing how Goldman Sachs derived more value from custom Excel additions than Microsoft made selling the software.
The host holds their own ▶ 45:11 Elena's antitrust historical parallelElena demonstrates high expertise by connecting cutting-edge MCP server architecture back to the Department of Justice's middleware arguments in 1990s Microsoft litigation.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The host as informed peer | Guest teaching | Guest disagreement | The host pushing back | Why |
|---|---|---|---|---|---|---|
| What Does Headless Software Really Mean? | 3 | 3 | 1 | 0 | Host Elena Burger guides the opening discussion on headless software and Salesforce's announcement, contributing a basic example about Slackbots while letting Seema explain the underlying data dynamics. | |
| Deconstructing Agents: Look Up, Act, and Analyze | 2 | 5 | 2 | 0 | Steven Sinofsky breaks down agents into a Look Up, Act, and Analyze matrix while humorously poking fun at industry jargon, while the host provides clean transitional questions. | |
| The Historical Sources of Software Stickiness | 6 | 4 | 1 | 2 | Elena demonstrates preparation by bringing up both guests' specific written work, citing Steven's essay rejecting the Sasspocalypse to frame the discussion on software stickiness. | |
| The Intricacies of Enterprise Logic and Customization | 1 | 6 | 1 | 0 | Guests exchange deep historical stories regarding enterprise logic, SAP customization, and Excel usage at Goldman Sachs, while the host remains largely silent during the narrative exchanges. | |
| Augmenting Legacy Software and Managing Enterprise Exceptions | 4 | 5 | 1 | 1 | Elena prompts the conversation around startups building on top of legacy SAP systems, leading to a discussion on exception handling and ad-hoc business processes. | |
| Long-Tail Workflows and Tapping Unstructured Data | 4 | 5 | 1 | 2 | Elena probes on whether automating the long tail is the primary bottleneck for developers and offers witty commentary on ambient data collection. | |
| The Middleware Fallacy and Strategic Enterprise AI Paths | 6 | 6 | 2 | 2 | Elena exhibits strong domain knowledge by drawing a historical parallel between modern MCP server middleware and 1990s Microsoft DOJ antitrust litigation. | |
| Startup Positioning and Internal Network Effects | 3 | 5 | 1 | 0 | Elena asks about startup positioning opportunities, prompting Steven to explain how startups should position themselves between established enterprise incumbents. |
Statements from this episode (25)
Sinofsky: People wildly underestimate the difficulty of vibe coding enterprise software
“There's this wild underestimation about, like, you could vibe code your way into enterprise software.”
Amble: Postgres and APIs cannot easily replace SAP's business logic
“Misconception right now is that you can just have, you know, Postgres database and APIs, and then bam, like you can replace SAP. That's like absolutely not true. That piece around the logic and everything else that is in captured in SAP is way, way more import…”
Sinofsky: People consistently fail to grasp exponential tech shifts in real time
“One of the things that happens in technology shifts is nobody understands exponential when it's happening.”
Amble: Software value shifts to data and logic in agentic architectures
“The UI doesn't matter because the agent isn't accessing the software via the UI. We could unpack whether the UI matters or not, but in the idea of it, that being headless is the data, the Logic, everything stored below it is really where the value is, not just…”
Amble: Salesforce Headless 360 is merely a rebranding of existing APIs
“I don't, from what I can tell, nothing actually changed. Their three 60 product was the same APIs that had always been exposed, now rebranded as their three 60 product.”
Amble: Notion users are more tech-savvy builders than Salesforce users
“Notion users tend, you know, all things being equal, are more, I'd say, tech-savvy and more agentic as builders”
Sinofsky: Current AI agents are just programs that take too long
“Like first you have this, you have a agent, which is far as I can tell right now is also a new word for program that takes a very long time to run and might not finish.”
Sinofsky: Most newly announced headless agent APIs are just lookup tools
“And in fact, many, many of the newly announced, you know, headless agent APIs are just look up. And they're just, you basically have a new interface to the old way to look something up.”
Amble: Incoming sales VPs routinely mandate adopting Salesforce
“When, like, new VPs of sales come into our companies, they, they're like, they mandate that Salesforce is there because they're used to using it, their teams are used to using it.”
Sinofsky: Collecting money is the stickiest enterprise software capability
“The most sticky thing you could do is actually collect money from a customer. And if you're collecting money, it turns out it's really, really hard for them to stop sending you money.”
Sinofsky: General Motors kept 600,000 Outlook seats due to calendar features
“General Motors isn't going to displace like 600,000 seats. Because of the calendar or some crazy thing like that.”
Sinofsky: The core software running Allstate is over 60 years old
“The software that runs all state is older than me. And it's not going anywhere.”
Sinofsky: Removing SAP would cause auto manufacturers and Walmart to evaporate
“If you take SAP out of a large automobile manufacturer, there's no automobile manufacturer left. Or Walmart. Like the company just evaporates. Because the company is defined, not just by purchasing the software, not even by just using it, but by how they codif…”
Sinofsky: Startups fail in enterprise software by designing for startup scale
“Startups look at enterprise software and they think about it in terms of startup scale... And that's fine until you have a 100,000 people in 20 countries with different national laws and policies about business expense and all of those, and then you overlay co…”
Sinofsky: Enterprise software's most used features are CSV and Excel exports
“In enterprise software, the two most frequently used features exist in no enterprise software natively. It's export to Excel and export as CSV and or PDF you pick.”
Sinofsky: Exception handling is the central challenge for AI agents
“This notion of exception handling is just the root of the challenge with agents, which is almost everything interesting in an enterprise is an exception.”
Sinofsky: Enterprises will trust AI to make autonomous decisions
“When we get through sort of the One point O version of this. We're going to get to a new version where people are like comfortable letting AI do or decide things because they realize it's adding a level of predictability and repeatability to their enterprise.”
Sinofsky: AI will make legal contracts longer and more sophisticated
“Except I can assure you contracts will get longer. And more sophisticated, and encompass way more sets of scenarios than a person ever could.”
Sinofsky: AI Is First Tech to Tap Unstructured Enterprise Data
“And I think that, that AI is the first thing to come along that really taps into that unstructured information in a company.”
Sinofsky: Middleware layers are structurally unstable in enterprise software
“This middleware layer, it's always, always very unstable. It looks great in a network hierarchy diagram of the OSI levels of networking, but it's just never that stable.”
Amble: Workday limits third-party API access and documentation
“Workday makes it extremely difficult to actually, like, get access to the documentation and work with the, like, and they don't expose all the endpoints.”
Amble: Legacy software incumbents will fail to build great AI agents
“I'm not bullish on the incumbent software building great agents on top.”
Sinofsky: Software startups should position between established players, not head-on
“The universal truth for enterprise software is the most difficult thing to do that happens to be the dumbest is to attempt to just compete head on with an existing category. And by head on, I mean, not just the same category, but doing it the same way. The big…”
Sinofsky: Incumbent software vendors will only bolt AI onto existing products
“The one thing that established players won't do is disturb their existing product line and go to market. So they absolutely will just be bolting AI on top of their existing product. They won't be getting rid of it. They won't stop working on it. They won't do …”