AI application layer disruption

Last week, Anthropic launched a legal automation plugin for Claude. Within hours, nearly $1 trillion was wiped from software and services stocks worldwide. Thomson Reuters dropped 18%. Relx plunged 14%. The FTSE 100, fresh off a record high, was dragged into the red.

The announcement raises a procurement question for IT leaders. Which application tasks can AI tools perform, and what would replacing an existing service require?

What actually happened

Anthropic announced a plugin for its Claude AI that could automate legal work: contract reviewing, NDA triage, compliance workflows, legal briefings, and templated responses. Alongside it came open-source tools for sales, customer support, and data analysis.

Shares fell across several information-services companies. Investors dumped shares in companies whose core business is packaging, indexing, and selling professional knowledge. Pearson fell 8%. Sage lost 10%. Wolters Kluwer shed 13% in Amsterdam. Experian dropped 7%.

Morgan Stanley called it "intensifying competition." William Blair flagged "a structural concern" about the relevancy and moats of information services companies under this new AI paradigm.

Software vendors need to explain what their products provide beyond tasks an AI model can perform.

The application layer shift

For the past two years, most enterprise AI discussions centred on infrastructure. Which cloud provider? What GPU allocation? How do we fine-tune a model? That was the AI conversation in 2024.

In 2026, the conversation has fundamentally changed.

Large language models are no longer just foundations that other companies build on. They're becoming the applications themselves. When Claude can review a contract, it's not augmenting Thomson Reuters' legal software. It can compete for some of the same work.

Moving into the application layer means AI providers sell tools for completing business tasks as well as models that other developers build on.

Why this matters for your software stack

If you're an IT Director or CTO running a typical enterprise, you've got 50 to 200 SaaS subscriptions. Many of those tools exist because they're good at one thing: taking unstructured data, applying rules, and presenting structured outputs. Contract management. Compliance tracking. Customer support routing. Sales intelligence.

That's precisely what LLMs do natively.

Not every SaaS tool is at immediate risk. Products with deep integrations, proprietary data moats, or network effects still have defensible positions. Tools that primarily provide search or automated workflows over general-purpose data face more direct competition.

Three strategic implications for IT leaders

1. Audit your stack for AI vulnerability

Take a hard look at every software contract coming up for renewal. For each tool, ask:

• What's the core value proposition? If it's "we make data easier to find and act on," that's increasingly commoditised by AI.
• Where's the proprietary data? Tools that generate unique, first-party data (your CRM, your ERP, your observability platform) are safer than tools that repackage third-party data.
• What's the switching cost? Deep integrations and workflow dependencies still create stickiness, even if the AI alternative is technically capable.

I'd recommend categorising your stack into three buckets: Safe (proprietary data, deep integrations), Watch (could be disrupted in 12-18 months), and At Risk (primarily knowledge-packaging or simple automation).

2. Renegotiate from a position of strength

Use renewal discussions to ask vendors how their products compare with emerging alternatives.

If you've got renewals coming up for legal tech, compliance tools, data analytics platforms, or customer support software, now is the time to push for better terms. A comparison can inform negotiations even when switching is not the right choice.

Practical moves:

• Request shorter contract terms (annual instead of multi-year)
• Push for AI-inclusive tiers at current pricing
• Ask vendors specifically what their AI integration roadmap looks like
• Negotiate exit clauses that account for "material capability change"

3. Build internal AI capability now

The companies that will benefit most from this shift are those with internal teams that can evaluate, implement, and govern AI tools. That means:

• An AI evaluation framework. When your legal team says "Can we just use Claude instead of our contract management tool?", you need a structured way to assess that, covering accuracy, compliance, data security, and total cost of ownership.
• Governance policies. Anthropic explicitly stated their legal plugin doesn't provide legal advice and outputs should be reviewed by licensed attorneys. Someone in your organisation needs to own these boundaries.
• Integration architecture. Build an integration layer where AI models can work alongside existing systems, pulling from your proprietary data and using the model's reasoning capabilities.

The vendor response will define winners and losers

Not every software company hit by the selloff is doomed. The smart ones are already adapting.

Thomson Reuters has been investing heavily in AI integration. Sage has built AI features into its accounting platform. Relx has been augmenting its research tools with LLM capabilities for over a year.

The vendors worth keeping are those treating AI as a feature, not a threat. They'll embed AI into their platforms, use their proprietary data as a differentiator, and offer capabilities that a standalone LLM simply cannot match, such as industry-specific compliance, audit trails and enterprise reliability.

The vendors to worry about are those still selling the same product they had in 2023, hoping their customers don't notice that a general-purpose AI can now do 80% of it.

What this means for the UK specifically

The UK is in a particularly interesting position. According to Morgan Stanley research, the UK is losing more jobs than it's creating as companies adopt AI, with a greater reported impact than the US, Japan, Germany, and Australia.

More than a quarter of UK workers are worried about job displacement from AI within five years. Clifford Chance, one of the world's largest law firms, has already cut London business services staff by 10%, citing AI as a factor.

For UK IT leaders, this creates a dual mandate:

1. Drive efficiency through AI adoption (your board is watching those stock movements too)
2. Manage the human impact thoughtfully (retraining, redeployment, and honest communication)

The organisations that handle this well will attract talent. The ones that don't will face retention crises as skilled workers flee to companies that seem to have a plan.

The practical takeaway

Here's what I'd do this week if I were sitting in an IT Director's chair:

1. Pull your SaaS renewal calendar. Flag everything renewing in the next 6 months.
2. Map each tool against AI capability. Be honest about what a model like Claude could feasibly replace.
3. Have the conversation with your CFO. Frame it as opportunity, not threat. "We could reduce our software spend by X% over 18 months while improving capability."
4. Start a pilot. Pick one low-risk workflow, such as internal document summarisation, first-pass contract review or support ticket categorisation, and test an AI-first approach.
5. Document everything. The governance, the results, the cost comparison. You'll need this when the board asks "what's our AI strategy?"

Looking ahead

The selloff is a reason to review vendor plans, not evidence that a specific replacement will meet your needs.

AI isn't replacing your entire stack overnight. But it is fundamentally changing the economics of knowledge work and the software built to support it. The IT leaders who recognise this shift early, negotiate smartly, and build internal capability will come out ahead.

Test candidate tools against your own workflows rather than relying only on vendor assurances.

For more on assessing AI tools in practice, see my posts on AI pragmatism in enterprise strategy and selecting AI tools for your business. If you need help rethinking your enterprise software strategy in light of AI disruption, my IT management consulting practice covers technology strategy, vendor evaluation, and portfolio rationalisation.

---

Daniel Glover is Head of IT Services at a UK e-commerce company, managing 50+ vendors and a multi-million pound technology budget. He writes about practical IT leadership and enterprise technology strategy at danieljamesglover.com.