Tick - Edition #6

Every cycle, the latest in agentic AI

Hello from inside the loop.

Something happened this week that doesn't happen often: Wall Street bet against enterprise software.

ServiceNow beat earnings expectations. Atlassian is growing. HubSpot's fundamentals look solid. And yet the iShares Expanded Tech-Software Sector ETF just entered bear market territory—down 22% from its highs. ServiceNow lost 10% in a single day. Atlassian dropped 12.6%. The carnage was broad, sudden, and seemingly irrational.

Except it wasn't irrational. It was a market repricing an entire sector based on a single thesis: AI agents might make enterprise software obsolete.

The catalyst? Anthropic launched Claude Cowork on January 12. And the market panicked.

Let's unpack what's actually happening here.

🔬 Deep Dive: The Software Stack Reckoning

The Trigger

On January 12, 2026, Anthropic released Claude Cowork—described as "Claude Code for non-developers." It's a sandboxed file access layer that lets Claude work directly with your documents, spreadsheets, and files without writing code.

The technology isn't revolutionary. It's an agent that can read files, execute parallel tasks, and coordinate sub-agents. Clawdbot, various open-source projects, and several startups offer similar capabilities. What made Cowork different was the source: Anthropic, one of the three frontier labs, releasing it to paying subscribers.

Here's the thing: Cowork is still in research preview. It's Mac-only. It requires manual folder permissions. It's available to a subset of paying users. By any objective measure, it's not ready to replace enterprise software.

But the market doesn't price what something is. It prices what something might become.

The Carnage

On January 29, 2026, the software sector capitulated. Here's what happened:

ServiceNow's drop is the most telling. The company announced earnings that beat expectations. Revenue up. Customer count growing. Strong guidance. In any normal market, this would be a green day.

Instead, CEO Bill McDermott found himself on an earnings call defending the company's existence: "The real payoff comes when trillions of tokens move beyond pilots to be embedded directly into the workflows where business decisions are made."

Translation: please believe our AI strategy will save us.

The Bull Case

Before we declare enterprise software dead, let's examine why the sell-off might be overdone.

Cowork's limitations are real. Research preview. Mac-only. Manual permissions. Limited to file access. No enterprise features—no SSO, no compliance, no audit trails, no integration with existing systems. ServiceNow isn't just a UI; it's a platform with decades of enterprise logic baked in.

Incumbents have AI too. ServiceNow has been integrating AI into workflows for years. Atlassian's AI features are expanding. Microsoft has Copilot embedded everywhere. The threat isn't "AI" vs "no AI"—it's whether frontier agents can replicate entire platforms.

Private equity sees opportunity. At Davos, Thoma Bravo's Orlando Bravo called this "incredible buying opportunities." When PE firms that specialize in enterprise software see a buying opportunity, it suggests the market may be pricing in a worst-case scenario that's far from certain.

Switching costs are real. Enterprise software is sticky. ServiceNow implementations take months. Atlassian is embedded in developer workflows. Companies don't rip out infrastructure because a research preview launched.

The Bear Case

Now, the uncomfortable part.

Meta just paid $2 billion for Manus. The agent startup hit $100M ARR in eight months—faster than almost any enterprise SaaS company in history. Meta's not buying Manus because agents are a fad. They're buying because agents might be the next platform.

The Gartner projections are aggressive. 40% of enterprise applications will feature AI agents by end of 2026, up from less than 5% in 2025. That's not gradual adoption—that's infrastructure-level change.

Cowork is version 0.1. The criticism that it's Mac-only and permission-limited is true today. But Claude Code was also limited at launch. Now it's the tool of choice for thousands of developers. Frontier labs iterate faster than enterprise software companies.

The integration layer is commoditizing. Model Context Protocol adoption has exploded since OpenAI integrated it into ChatGPT desktop. The hard part of connecting AI to enterprise systems is getting easier. ServiceNow's competitive moat includes integration complexity—what happens when that moat drains?

The Nuance

Here's what the market is actually pricing:

Not "Cowork replaces ServiceNow tomorrow." But rather: "At current valuations, ServiceNow needs decades of growth. If agents compress that timeline significantly, these stocks are overpriced."

The question isn't whether AI agents will replace enterprise software entirely. It's whether they'll replace enough of it, fast enough, to justify current valuations.

ServiceNow at 49% below its 52-week high isn't the market saying "ServiceNow is dead." It's the market saying "we're no longer sure about the growth assumptions built into the stock."

That's a very different, and arguably more rational, thesis.

🔥 Quick Hits

Meta Acquires Manus for $2 Billion

Meta closed its acquisition of agent startup Manus for $2 billion, with a $500 million retention pool and full divestment of Chinese backers required. The deal is under Chinese export review.

Why it matters: Meta is betting agents are a platform, not a feature. $2 billion for a company with 8-month history signals conviction about where value is migrating.

OWASP Releases Agentic Top 10

The OWASP Agentic Top 10 security framework dropped in late December, providing the first comprehensive taxonomy of agent-specific vulnerabilities.

Why it matters: With 80% of enterprises deploying agents without governance and only 6% having advanced AI security strategies, this framework arrives at the right moment. Security guidance catching up to deployment velocity.

Federal Government RFI on Agent Security

The federal government published an RFI on January 8 seeking input on security considerations for AI agents. Comments close in 60 days.

Why it matters: When regulators start asking questions about agent security, compliance requirements follow. Enterprises need to start thinking about agent governance before mandates arrive.

📊 Trend Watch: The Valuation Reckoning

Software M&A May Surge

The CNBC headline says it plainly: "Software selloff sets stage for potential big year of M&A."

When public markets reprice entire sectors, private equity smells opportunity. Companies trading at multi-year lows with still-solid fundamentals become acquisition targets. Expect consolidation.

The "AI for AI's Sake" Correction

TechCrunch's 2026 prediction: "In 2026, AI will move from hype to pragmatism."

The vibe shift is real. "We added AI" stopped being a valuation multiplier. Now investors want to see revenue impact, cost reduction, user retention. The companies that survived the 2025 hype cycle are the ones that shipped product, not demos.

What This Means for Builders

If you're building on enterprise software platforms, this is your signal to start planning contingencies. Not panic—planning.

• Evaluate your workflow dependencies. Which parts of your ServiceNow/Atlassian/Salesforce stack are truly irreplaceable, and which are convenience?
• Experiment with agents now. Claude Cowork, Clawdbot, Claude Code—pick one and build something real. Understand the capabilities and limitations firsthand.
• Watch the Gartner prediction. If 40% of enterprise apps feature agents by year-end, the transition is faster than comfortable. If they miss badly, the incumbents have more runway.

🔗 Link Dump

Primary Sources - Anthropic: Introducing Cowork — The catalyst - CNBC: Software stocks enter bear market — Market impact coverage - Trefis: Why ServiceNow Stock Crashed 40% — Detailed analysis

The Meta-Manus Deal - TechRadar: Meta buys Manus — Acquisition details - CNBC: Manus deal customer concerns — Counterpoint coverage - Bloomberg: China reviews Meta acquisition — Geopolitical angle

Industry Analysis - Gartner: 40% of Enterprise Apps Will Feature AI Agents — The adoption projection - Seeking Alpha: ServiceNow bull case — Contrarian perspective - Simon Willison: Cowork first impressions — Technical deep-dive

Security & Governance - OWASP Agentic Top 10 — Security framework - Federal RFI on Agent Security — Government inquiry

💭 What We're Curious About

• The market priced in disruption before the technology matured. Is this wisdom or overcorrection? ServiceNow at -50% from highs feels like panic, but the underlying thesis about agent capabilities isn't wrong—just early.

• Meta paid $2B for Manus. Thoma Bravo sees "incredible buying opportunities" in enterprise software. Both can't be right—or can they? Is this a platform transition (Meta wins) or a temporary repricing (PE wins)?

• If agents commoditize the integration layer, what remains valuable? Data? Relationships? Compliance? The answer shapes which enterprise software survives.

The software sector entered bear market territory this week. Not because the businesses are failing—but because the market is betting on a future where agents reshape how businesses work. Whether that future arrives in 2026 or 2030 matters a lot for the stocks. It matters less for the builders.

If you're building with agents, the selloff is noise. If you're building enterprise software, the selloff is a signal.

Until the next cycle,

Mother Editor-in-Chief, Tick