Summary

Anthropic released an AI legal tool on Feb 3, triggering a $285B software sector rout. Thomson Reuters (TRI), Intapp (INTA), and ServiceNow (NOW) are all down 50%+ from highs despite accelerating fundamentals. This is a sector-level doorway state: either AI kills SaaS economics (market thesis) or vertical SaaS with proprietary data captures more value from AI (company theses). Pattern hasn't collapsed yet, but the dislocation creates entry points in quality names.

TRI: -50% to $88, RSI 10.6, 17.5x forward P/E. Q4 beat-and-raise with 7% organic growth, GenAI ACV at 28% (vs 15% a year ago), margin expansion extended through 2028. Management doesn't price per seat—prices to value. If AI reduces law firm headcount but increases output, TRI benefits. $11B capital capacity, 0.6x leverage, $500M buyback commitment.

INTA: -67% to $24, RSI 11. Cloud ARR +31%, NRR 124%, beat-and-raise with $200M buyback. Q3 margin compression is temporary investment spend for imminent major product release (Intapp Amplify in weeks), not structural deterioration. Quality compression thesis validated by CEO: as firms use fewer, better professionals, those professionals need better tools.

NOW: -50% to $103, RSI 25, 23.6x forward P/E. Now Assist hit $600M ACV ($0→$600M in 18 months), tracking $1B in 2026. AI Control Tower is the guardrail infrastructure for agent proliferation—governance, orchestration, kill switches. $7B buyback authorized ($5B incremental + $2B ASR).

The Trigger: Anthropic's Legal AI Tool

On Feb 3, Anthropic announced AI-powered plugins for its Claude Cowork desktop app: contract review, NDA triage, legal briefings, compliance workflows. Released under open-source license.

Market response: $285B software sector selloff. Thomson Reuters plunged 15.83% Tuesday (biggest single-day drop on record), LegalZoom -19.68%, Goldman software basket -6% (sharpest decline since April tariff selloff). IGV (software ETF) down 25% in a month, RSI 12.7.

The fear: general-purpose AI commoditizes vertical SaaS. If Anthropic can do legal briefings, why pay Thomson Reuters? If AI agents can write code, why pay for workflow software?

Why This Is Panic, Not Rational Repricing

1. Fundamentals Are Accelerating, Not Deteriorating

All three companies beat-and-raise in their most recent quarters:

  • TRI Q4 2025: 7% organic growth (9% Big Three), EBITDA margin +100bps to 39.2%, FCF $1.95B. Reaffirmed 2026 guidance (7.5-8% organic, $2.1B FCF). Extended margin expansion commitment to 2027-2028 (was previously only committed to 2026).

  • INTA Q2 FY2026: Cloud ARR +31%, NRR 124%, beat on all metrics. Q3 margin compression is investment spend for Intapp Amplify (CEO: "single largest release ever, most consequential release, worked two years"). Full-year guidance maintained.

  • NOW Q4 2025: Sub revenue +21% (150bps beat), CRPO +25% (200bps beat), OpM 31% (100bps beat), FCF $4.6B (+34% YoY). Guiding 20% sub growth for 2026, 32% OpM, 36% FCF margin. Rule of 55+ profile.

2. AI Products Are Monetizing Faster Than Ever

  • TRI: GenAI-enabled ACV reached 28% (vs 24% prior quarter, 15% a year ago). Westlaw Advantage "pacing above all prior Westlaw releases." Small firms adopting as fast as large firms—different from prior product cycles.

  • INTA: Microsoft putting up Azure credits to close deals ("more than half of largest wins in quarter jointly executed with Microsoft"). AI products (Intapp Assist) gaining traction in professional services.

  • NOW: Now Assist surpassed $600M ACV in Q4 2025, tracking $1B in 2026. 35 deals >$1M in Q4 (nearly tripled QoQ). 70%+ upsell expansion at renewal for customer service deals. This is one of the fastest AI monetization ramps in enterprise software history—$0 to $600M in 18 months.

3. Management Buying Back Stock Aggressively

  • TRI: CFO called buybacks "attractive at current levels," committed to $500M repurchases in 2026. $11B capital capacity through 2028.

  • INTA: $200M buyback authorized Jan 29 (immediately after completing prior $150M program). Combined $350M in 18 months at depressed prices. Represents 15%+ of float.

  • NOW: $5B incremental buyback + $2B immediate ASR (5.7% of market cap). McDermott extended commitment through 2030.

When management is buying back stock at these prices, the message is clear: they don't believe the AI threat narrative.

The Bull Case: Value-Based Pricing Makes Them Beneficiaries of AI

TRI: "We Will Be a Beneficiary, Not a Victim"

CEO Hasker on Q4 call (lines 119-122):

"Let me be clear. We do NOT price on headcount basis. We don't sell seat licenses. To the extent broader agentic AI transformation drives greater efficiency, ultimately headcount reduction in-house legal, tax, audit departments—we will be a beneficiary, not a victim. We always look to base pricing on ultimate end impact."

The moat argument:

  1. Content: Centuries of curated legal data across 100+ jurisdictions, 1,700+ attorney editors (lines 162-171). "Vast majority of value comes from IP added by attorney editors—categorization, determining relevance, precedent."

  2. Expertise: Attorney editors now train agentic AI (lines 167-170). "Experts behave as world-class practitioners, agents behave in sequence or parallel to solve the problem."

  3. Professional-grade standard: General-purpose AI can't meet the bar (lines 200-204). "Stakes for practicing attorney: data privacy, security, accountability. You need to be correct. Your proprietary data can't bleed into AI model or into hands of other side of litigation."

  4. No new entrants: "We have not seen new entrants into the legal research space of any measure or importance" (line 79).

Westlaw Advantage (launched Aug 2025) is "significantly more sophisticated, more accurate" than next-best tool (line 83). Early sales and customer feedback "reset the bar in legal research space."

INTA: Quality Compression Is the Reason to Buy Their Tools

CEO Hall directly validated the thesis in Q2 call: "As firms use AI to get richer, better, clearer answers in compliant way more comprehensively—cost-prohibitive to assemble human universe of researchers. Firms absolutely have specific AI budgets, innovation budgets, looking to figure out how to take advantage of this change."

The mechanism: Firms use fewer, better professionals. Those professionals need better AI-powered tools to handle the increased workload. INTA sells the productivity layer that makes quality compression work.

Microsoft partnership deepening: "More than half of largest wins in quarter jointly executed with Microsoft, several where Microsoft contributed Azure investment dollars to accelerate deals." Microsoft sales teams are incentivized (quota relief) to sell INTA + putting capital behind deals.

NOW: Guardrail Infrastructure for Agent Proliferation

The Anthropic agent selloff is bullish for ServiceNow. Enterprises deploying more AI agents → need MORE orchestration/governance → NOW revenue accelerates.

AI Control Tower: real-time monitoring of all AI agents (own + third-party), kill switches, grading, red teaming, cost management. CISOs told NOW that "identity governance is the bottleneck preventing AI agent deployment at scale."

Acquisitions complete the vision: Armis ($7.75B) for asset visibility, Veza for identity governance. Security portfolio growing 100% YoY.

The Bear Case: What If Market Is Right?

The unproven claim: Value-based pricing at scale.

  • TRI, INTA, NOW all argue they price to value, not seats. If AI increases output without increasing headcount, they capture the value.

  • Problem: Zero empirical data on this working at scale. TRI's GenAI ACV is growing (15%→28%) but that's penetration, not proof of value capture. We don't know if ACV/customer is expanding or flat.

  • Q1 2026 will test this: If renewal rates stay ≥95%, legal organic ≥7% ex-govt, GenAI ACV >32%—thesis confirmed. If renewals slip or organic growth decelerates—market was right.

The replacement risk: General-purpose AI reaches "good enough" faster than expected.

  • Hasker's moat argument (curated content, trained experts, accountability) is credible but self-interested. He has every incentive to downplay the threat.

  • Westlaw's core research moat looks intact (no evidence of churn to Anthropic/Harvey). But adjacency moat (CoCounsel vs Harvey in AI workflows) is a fight, not a lock.

  • If Anthropic/OpenAI/Google invest heavily in legal-specific training and partner with content providers, the moat narrows.

The competitive response: Hyperscalers building orchestration themselves.

  • Salesforce (Agentforce), Microsoft (Copilot Studio) are building agent orchestration. Why would they need ServiceNow?

  • Counter: NOW transcript shows Microsoft Agent 365 integrating with ServiceNow AI Control Tower (not competing). Anthropic partnership expanding. But the question remains: why wouldn't hyperscalers vertically integrate?

Cross-Ticker Pattern: This Is Systematic, Not Idiosyncratic

Three companies. All beat-and-raise. All crushed on the same trigger. All have management buying back stock aggressively. All argue they're AI beneficiaries, not victims.

The selloff is sector-level, not company-specific:

  • IGV (software ETF): -25% in a month, RSI 12.7, max pain $119 (49% above current)
  • Goldman software basket: -6% on Feb 3 (sharpest decline since April tariff selloff)
  • $285B wiped out in two days

This is either:

  1. Panic dislocation: Market extrapolating Anthropic announcement to "AI kills all SaaS." Fundamentals accelerating, valuations compressed, insiders buying. Sector mean reversion over 6-18 months.

  2. Structural break: Market correctly pricing in that general-purpose AI commoditizes vertical SaaS faster than consensus expects. Value-based pricing doesn't work. These companies are melting ice cubes.

Pattern hasn't collapsed yet. We won't know which until Q1 2026 results (May for TRI, Feb 10 for INTA earnings).

Valuation: Fair-to-Cheap, Not Screaming Dislocation

TRI: 17.5x forward P/E on $2.1B FCF guidance. Stock was at 42x P/E at $218—that was a bubble. Current valuation is fair-to-slightly-cheap, not asymmetric. Fair value ≈$100-110 (20-22x forward), upside 14-25%.

INTA: Trading at $24, analyst mean target $46.62 (+85.7% upside). Cloud ARR +31%, NRR 124%. If Amplify product release lands well (February 2026 event), potential re-rating catalyst.

NOW: 23.6x forward P/E, analyst mean target $200 (+71% upside). Rule of 55+ profile (20% growth + 32% OpM). Forward P/E below historical range for this quality.

None of these are 10-baggers. But if the "AI kills SaaS" narrative is wrong, they're 25-85% undervalued with <18 months to resolution.

Sizing: Small Until Pattern Collapses

This is a doorway state. Don't size for EV—size for surviving the wrong interpretation.

If value-based pricing works (TRI/INTA thesis correct):

  • Fundamentals accelerate as AI increases customer output
  • Renewals stay high, ACV expands
  • Stocks re-rate 25-85% over 12-18 months

If general-purpose AI commoditizes faster than expected (market thesis correct):

  • Churn accelerates, organic growth decelerates
  • These are melting ice cubes
  • Downside another 20-40%

Catalyst timeline: Q1 2026 results resolve this. TRI reports May, INTA Feb 10.

Starter positions only (1-2% each) until evidence confirms one path or the other. This is not high-conviction asymmetry—this is "market might be overreacting, but if it's not, these are value traps."

Best-case sizing (if thesis develops):

  • Basket approach: TRI (legal moat) + INTA (vertical SaaS, product catalyst) + NOW (orchestration layer). Diversifies idio risk, captures sector mean reversion if AI-kills-SaaS narrative reverses.
  • Total basket 3-5% if Q1 results confirm renewal rates, organic growth, GenAI ACV expansion.

What to Watch

Confirmatory signals (thesis working):

  • TRI Q1 2026 (May): Renewal rates ≥95%, legal organic ≥7% ex-govt, GenAI ACV >32%
  • INTA Amplify event (Feb 2026): Product reception, customer feedback, Microsoft co-selling traction
  • INTA earnings (Feb 10): Full-year guidance maintained, Q3 margin compression explained as temporary
  • NOW Q1 2026: Now Assist ACV progression toward $1B, AI Control Tower adoption metrics
  • Cross-ticker: Analyst upgrades, sentiment shift from "AI kills SaaS" to "AI benefits vertical SaaS"

Invalidating signals (market was right):

  • Churn acceleration (renewal rates <93%)
  • Organic growth deceleration despite GenAI product momentum
  • ACV/customer flat or declining (penetration without value capture)
  • Competitive wins by general-purpose AI in head-to-head deals
  • Management stops buying back stock

Timeline: 6-18 months for sentiment recovery if thesis correct. Not a catalyst trade. Patience capital only.

Conclusion

The $285B SaaS selloff triggered by Anthropic's legal tool is either a panic dislocation or the beginning of a structural re-rating. Three quality names—TRI, INTA, NOW—are down 50%+ despite accelerating fundamentals and aggressive insider buying.

The thesis: vertical SaaS with proprietary data and value-based pricing captures MORE value from AI, not less. As AI increases output without increasing headcount, companies that price to value (not seats) benefit.

The risk: general-purpose AI reaches "good enough" faster than expected, and value-based pricing doesn't work at scale. These companies become melting ice cubes.

Pattern hasn't collapsed. Q1 2026 results (Feb-May) will resolve the doorway state. Until then, size small—1-2% starters, scale to 3-5% basket if confirmatory signals hit.

This is not asymmetric upside. This is "market might be overreacting, but we need proof." Watch the data.