The Anthropic Impact and Market Reaction: B2B Information Services Faces Evolution, Not Extinction

In early February, shares of major business information firms plummeted sharply as markets reacted to a perceived AI inflection point. Anthropic unveiled agentic capabilities in its Claude assistant positioning it as a system that can automate white-collar tasks. This triggered a broad sell-off in data and software companies.

Thomson Reuters (owners of Westlaw legal database) fell nearly 18% in one day, while peers RELX (parent of LexisNexis) and Wolters Kluwer dropped 14% and 13% respectively.

Other data, analytics and workflow software firms were dragged down alongside them.

The reaction is reflective of investor fears that AI advances could disrupt the economics of the Business Information services sector.

A core concern is that as AI agents take on workflows previously conducted within proprietary platforms, control of the user interface may shift away from incumbent providers. If orchestration resides elsewhere, these firms risk being repositioned as upstream data suppliers and their ability to capture value reduces. The durability of sticky seat-based subscription models – long part of the sector’s appeal to investors – will also be challenged.

But critically this technological development also increases the need for authoritative intelligence in high-stakes decisions. Agentic systems require reliable, auditable and defensible inputs to function well — attributes that sit at the core of the historical moats built by leading B2B information firms and have supported their sustained above-market growth.

The sector is not facing extinction. But not all businesses are guaranteed to continue to succeed.

Winners will be those that deepen proprietary data advantages, embed themselves deeper into mission-critical decisions, evolve commercial models and integrate AI into their own products and processes. Others may see their role — and economics — narrow.

What changed and why investors reacted

Claude’s new capabilities did not materially improve search or summarisation. They signal a shift from assistive to agentic AI.

In simple practical terms, agentic systems can:

• Plan multi-step tasks
• Invoke tools and APIs
• Apply rules-based playbooks
• Execute workflows autonomously
• Produce structured decision-ready outputs

In the example of Cowork legal plug-in – the most visible catalyst for the sector’s repricing – that means automated clause-level contract review, document triage, compliance checks and structured briefing generation.

Similar capabilities are emerging across other white-collar domains.

The market reaction reflects three core concerns

1. Workflow displacement risk & value capture migration 
If foundation model providers move to workflow orchestration, they sit above traditional data platforms and tools. The AI agent becomes the primary interface.

Consider a law firm using an AI agent for contract review. The agent pulls from a third-party precedent database and internal documents. The lawyer interacts with the output from the AI interface, not directly with the database.

Even if the third-party database remains essential, it is no longer perceived as the platform. The AI layer controls the workflow and the user relationship

The risk is not the irrelevance of the data. It is loss of control over the execution layer where economic power sits

2. Commercial model fragility 
Many B2B information services businesses rely heavily on seat-based subscriptions. More users means more revenue.

Agentic AI weakens that anchor.

If one senior lawyer now oversees reviews that previously required multiple associates – each with a paid seat –  procurement teams start to ask ‘why are we paying per user? Should pricing reflect output instead?’

Even with stable demand, pricing logic will shift from headcount to usage or workflow value. Markets are discounting whether commercial models can evolve without eroding margin

 3. Fundamental information substitution risk 
If data is not proprietary or unique, AI may pull from cheaper competitors or approximate from public or aggregated sources.

A regulatory intelligence provider that just aggregates and summarises public updates, for example, may find that an agent can scan those same sources directly and generate similar outputs.

Where information differentiation is shallow, AI dramatically increases the risk of substitution and compresses value.

This is a sorting event.

The repricing does not imply that quality data businesses will become redundant. Far from it. As automation increases, the likelihood of error rises. The need for reliable, structured information in high-stakes workflows becomes more important, not less.

What we should be reassessing is which firms genuinely possess those attributes.

The implications extend beyond large public incumbents. The same structural questions apply across the broader B2B information landscape – from financial market data providers and credit risk platforms to specialised vertically focussed data intelligence firms.

For most, some degree of evolution will be required.

What the winners will look like

The B2B Information services business that emerge successful will share five characteristics:

1. They sit in high-stakes workflows and own the information that matters.

They operate in areas where errors carry material legal, financial or operational consequences.

They are recognised as an authoritative, differentiated source where quality materially increases the probability that a decision is correct.

Defensibility does not require every raw input to be truly proprietary. Value can stem from proprietary models and scoring systems, structured interpretation and analytical layers, decision frameworks, longitudinal data or unique taxonomies and identifiers

The acid test is simple: if an agent does not use the information, will it be more likely to make an important mistake? If not, the position is vulnerable.

2. They choose their role in position in their stack deliberately 
Simplistically there are two defensible positions. Many firms will blend them.

a. Indispensable data input
AI agents and tools must call a firm’s information as the authoritative source to function correctly. Commercial relationships reflect this dependency.

b. Specialised AI agent for specific high-stakes use case
Firms control high-stakes, domain-specific AI tools

They embed AI into front-end tools in ways that improve client productivity and usability and deepen dependence.

3. They structure information for agent consumption 
Winning firms understand that the primary consumer of professional information will increasingly be an AI system.

They make their data easy for agents to integrate: machine-readable and structured, accessible through clean APIs, accompanied by strong provenance trails and version controls.

They are easy to integrate, but hard to replicate.

4. They treat AI licencing as a strategic lever 
Licencing decisions will shape competitive positioning. The objective is not just to ‘integrate with AI’ but” but to determine where durable control, differentiation and pricing power can sit.

Licencing can either reinforce structural dependence on information or accelerate disintermediation

5. They use AI to strengthen the information engine itself
Winners will not only respond to how AI is changing their distribution strategy and front-end tools. They will deploy it aggressively in their open operations, including in their data pipeline.

They will use AI to improve how data is sourced, extracted, structured, normalised and quality-controlled.

Executed successfully, this strengthens the data moat: better coverage, greater consistency and faster updates. And it lower costs and improves margins

In conclusion

This is not about AI replacing vast swathes of B2B information businesses. It is about clarifying which have the characteristics to succeed and which do not.

The public market repricing reflects investor uncertainty about differentiation, commercial model resilience and workflow control.

In an agentic world, durable advantage will accrue to businesses that control differentiated, authoritative intelligence, are embedded in critical workflows and are proactive and strategic in determining how and where their information is consumed.