Software's new model

---

1. Key Themes

Theme 1: Agentic AI Is Shifting Software from Tools to Outcomes — And Capital Is Flooding In

The dominant story is a structural change in how AI software is sold, valued, and defended. The pitch is no longer "use our tool" but "buy our workflow."

"VC-backed companies in agentic AI raised over $24 billion across more than 1,300 deals—equivalent to 73% of all dollars in the space from 2015 through 2024... The rapid rise means we've turned the corner from experimentation to deployment."

"Agentic systems execute end-to-end workflows rather than assisting with individual tasks, which lets providers charge for results, not just usage. That will change how software is built, sold and valued."

Capital is also concentrating within the sector: "Nearly 85% of agentic AI companies operate in IT... Cybersecurity, developer tooling and enterprise productivity are dominating because they can be deployed quickly and their ROI is easy to measure."

---

Theme 2: Model Performance Is Becoming a Commodity — Moats Are Organizational, Not Technical

A critical signal for both investors and builders: the source of competitive advantage is moving away from the underlying model.

"Adoption isn't about how good the model is anymore. Across industries, founders and operators pointed to governance, organizational readiness and integration depth as the real limiting factors."

"On defensibility, almost every company described their moat in terms of proprietary organizational context, workflow embedding and switching costs that compound over time. Model performance is expected to commoditize."

This suggests that durable agentic AI businesses will win through depth of integration and data lock-in, not foundational model superiority.

---

Theme 3: Trust and Reversibility Are the New Adoption Gatekeepers

The article surfaces a nuanced but critical insight about where AI adoption actually stalls — and it isn't capability.

"Trust is increasingly an engineering problem. Respondents across sectors independently identified explainability, audit trails and the reversibility of agent actions as prerequisites for expanding their autonomy in businesses. Adoption is furthest along where mistakes are recoverable, and stalls where they are not."

This has real implications for enterprise sales: products that build in auditability and "undo" mechanics will penetrate regulated or high-stakes environments faster.

---

Theme 4: Healthcare PE Is Outperforming in a Brutal Fundraising Market

While the broader PE fundraising market cratered, healthcare-specialist funds proved to be a standout exception — with both stronger close rates and superior returns.

"The amount of capital [healthcare specialist managers] raised fell by 4.2% in 2025, from $19.1 billion to $18.3 billion, compared with a 31.2% drop for all global PE funds."

"75.8% of healthcare PE funds hit their target size, and 36.4% exceeded it. During the same period, 72.6% of generalist, middle-market PE funds met their targets, and 26.1% exceeded them."

"Top-quartile healthcare PE funds have consistently outperformed the wider PE market across vintages from 2014 to 2020."

---

Theme 5: Smaller Agentic AI Players Face a Binary Fate — Scale or Be Acquired

The article ends with a pointed prediction about market structure in agentic AI.

"Investors will pile into companies that can orchestrate, integrate and operate across systems at scale. Those that can't are likely to be bought by bigger platforms."

This is a consolidation signal: mid-tier agentic AI startups without deep workflow integration are acquisition targets, not independent durables.

---

2. Contrarian Perspectives

The "SaaSpocalypse" May Be Geographically Contained

While concern about AI disrupting SaaS business models (dubbed the "SaaSpocalypse") is widespread in US private credit markets, non-US lenders aren't buying the panic.

"These investors insist the SaaSpocalypse is just an American problem. Some non-US lenders, especially in Europe, aren't worried about the state of the private credit market."

This geographic divergence in sentiment could represent a mispricing opportunity — either US lenders are overreacting, or European lenders are underestimating the contagion risk.

---

VC Secondaries Can Be a Liability, Not Just a Liquidity Mechanism

The conventional view is that secondary markets provide welcome liquidity for employees and early investors. But Anthropic's failed tender offer — where employees held onto more shares than expected — illustrates that secondary access can actually signal concern or dysfunction.

"Anthropic falls short in its tender offer, with employees holding on to more shares than expected... Read more about how venture secondary deals can be a liability."

If employees are declining to sell at presumably attractive prices, it may reflect either strong long-term conviction or, alternatively, a lack of external market confidence creating a standoff.

---

Private Credit Is Not a Bubble — Despite Market Anxiety

Against a noisy backdrop of private credit fears, at least one senior practitioner is pushing back forcefully.

"While private credit and software concerns have rattled markets, the asset class isn't a bubble or likely to cause the next economic downturn, said David Gross, managing partner at Bain Capital."

This is a direct counter-consensus view from a sophisticated operator — worth weighing against the prevailing bearish narrative on private credit leverage.

---

3. Companies Identified

---

4. People Identified

---

5. Operating Insights

Build for Reversibility Before Selling to Enterprises

The single most actionable insight from the agentic AI founder interviews is about sequencing adoption — don't start with high-stakes, irreversible workflows.

"Adoption is furthest along where mistakes are recoverable, and stalls where they are not."

For operators: prioritize use cases where agent errors can be caught and corrected (e.g., drafting, logging, routing) before pushing into autonomous decision-making in regulated, financial, or clinical contexts. Design audit trails and rollback mechanics as core product features, not afterthoughts.

---

Moat-Building in Agentic AI Is About Integration Depth, Not Model Quality

Founders and operators should stop competing on model benchmarks and instead focus on embedding deeply into client workflows and accumulating proprietary organizational context.

"Almost every company described their moat in terms of proprietary organizational context, workflow embedding and switching costs that compound over time. Model performance is expected to commoditize."

The practical implication: prioritize long implementation cycles, data sharing agreements, and workflow co-design with early customers over speed-to-market with a generic product.

---

Healthcare PE Is a Defensive Allocation with Offensive Returns

For fund allocators and GPs, the healthcare PE data makes a compelling case for specialist over generalist exposure in a constrained LP environment.

"75.8% of healthcare PE funds hit their target size, and 36.4% exceeded it... Top-quartile healthcare PE funds have consistently outperformed the wider PE market across vintages from 2014 to 2020."

Middle-market healthcare (not mega-fund) is where the hit rate is highest and targets are being exceeded — a structurally attractive segment for both GPs raising and LPs deploying.

---

6. Overlooked Insights

The HumanX Signal: AI Backlash Is Real but May Be Obscuring Creative Sector Opportunity

Briefly mentioned but not developed, startup executives at HumanX flagged that public backlash against AI in the creative arts is potentially masking genuine unlocks.

"Startup executives at the HumanX conference said the backlash is real, but it's obscuring what AI can actually unlock for the creative arts."

This is worth tracking: if the public narrative around AI-generated content is causing investors to avoid creative AI startups, it may be creating a valuation discount in a sector with real product-market fit waiting to emerge.

---

The Upper Middle Class Is Now the Largest US Income Cohort — A Consumer Market Shift

A brief citation from a Morning Brew summary of an American Enterprise Institute report surfaces a macro demand-side shift with implications for consumer startups, premium product positioning, and wealth management platforms.

"A new report says the upper middle class is now the largest income group in the US. An American Enterprise Institute analysis says that more Americans got richer, shrinking the lower rungs of the middle class."

For entrepreneurs targeting consumer markets: the addressable market for premium, differentiated products and services may be structurally larger than historical models assumed.