Collage of Claude Sonnet 4.5 demo screenshots showing Imagine with Claude interface, managing agent context, Chrome extension workflow, and document creation features

The Demo Videos That Actually Changed the Conversation

Silicon Valley finally cracked the code. For years, AI companies have been drowning audiences in benchmark scores and vague promises. However, Anthropic’s Claude Sonnet 4.5 launch brought something refreshingly different: demos that actually demonstrate.anthropic+1

Instead of telling us what their AI can do, they showed us Claude playing Catan for 75 minutes, creating software in real-time, and organising messy home budgets across browser tabs. Additionally, they showed Claude creating polished documents ready for download and charting its own growth through different model versions.

Yet here’s the twist—after looking at each video from digital marketing, UX writing, AI advocacy, and end-user perspectives, the picture becomes more complex. Although these demos represent a big leap in AI marketing sophistication, they also reveal missed opportunities that could have made this launch genuinely game-changing.

The Enterprise Validation That Changes Everything

But first, let’s acknowledge what Anthropic got spectacularly right. Rather than just launching a model, they delivered a complete enterprise validation story. Furthermore, this included customer testimonials that would make any B2B marketer weep with joy, an open SDK that makes agent development easier , and safety credentials that position them as the grown-ups in an industry still finding its feet.fastmcp+1

The Complete Demo Portfolio: Five Strategic Touchpoints

Anthropic’s launch included five distinct demonstrations. Moreover, each one targeted different aspects of professional workflow integration:

1. Context Management Through Gaming

First, Claude playing Catan for 75+ minutes, building opponent profiles whilst automatically removing irrelevant information.

2. Real-Time Software Creation

Next, “Imagine with Claude” creating interfaces on demand without writing code.

3. Browser Integration

Then, Chrome extension managing complex workflows across multiple platforms.

4. Document Creation

Also, generating polished files (docs, presentations, spreadsheets) ready for immediate usereuters

5. Capability Evolution

Finally, showing progressive improvement across model versions through actual performanceeconomictimes

This comprehensive approach addressed the complete spectrum of professional AI applications. Specifically, it covered strategic thinking, creative generation, workflow automation, and document production.

Additionally, the timing was exquisite: dropped just before competitors’ announcements, forcing the market to respond to Anthropic’s narrative rather than vice versa.

The Enterprise Proof Points That Actually Matter

Customer Testimonials That Aren’t Just Marketing Fluff

Rather than using anonymous “beta testers,” Anthropic played the credibility game with surgical precision : datastudios+1

GitHub Copilot claims “exceptional edits” with error rates dropping from 9% to 0%. Moreover, they report improved multi-step reasoning capabilities.
Canva highlights Claude’s impact on 240 million users through enhanced design capabilities. Furthermore, they report faster feature development.
Figma reports improved prototyping experiences. Additionally, they note smoother user interactions in early testing.
Security teams saw Claude decrease vulnerability review time by 44%. At the same time, they increased accuracy by 25%—metrics that make CISOs pay attention.

Real Enterprise Success Stories

Bridgewater Associates, the world’s largest hedge fund, deployed Claude Opus 4 through Amazon Bedrock for investment research. As a result, they achieved “first-year analyst-level precision” whilst reducing time-to-insight by 50-70%.datastudios

TELUS scaled Claude across 57,000 employees, creating 13,000+ AI-powered tools. Consequently, they saved over 500,000 staff hours. This resulted in $90 million in measurable business benefit.datastudios

The Claude Agent SDK: Platform Play Disguised as Developer Tool

The Claude Agent SDK represents Anthropic’s most underreported strategic move. By open-sourcing the infrastructure that powers Claude Code, they’ve essentially created a developer ecosystem around their models. However, this isn’t just good karma—it’s horizontal scaling through community adoption.anthropic+1

JetBrains immediately integrated the SDK into their IDEs. Meanwhile, Zapier deployed over 800 internal Claude-driven agents with 89% employee adoption. The SDK includes automatic context management, rich tool ecosystems, advanced permissions, and production essentials. Therefore, it gives developers everything they need to build sophisticated agents without reinventing infrastructure.jetbrains+1

The Digital Marketing Perspective: Strategic Brilliance with Execution Gaps

What Worked Like Swiss Clockwork

Target Audience Precision

Target Audience Precision was Anthropic’s secret weapon. Instead of trying to appeal to everyone, they laser-focused on developers and enterprise buyers. These are people who actually understand the difference between a 30-hour autonomous coding session and typical AI assistance. Furthermore, this wasn’t demographic targeting—it was psychological precision.techcrunch+1

Show, Don’t Tell Strategy

The “Show, Don’t Tell” strategy deserves marketing textbook inclusion. Rather than another PowerPoint about “revolutionary capabilities,” they demonstrated Claude playing strategy games, managing complex workflows, building software on the fly, and generating downloadable documents. As a result, each video answered the core buyer question: “Will this actually work for my use case?”

Customer-Led Validation

Customer-Led Validation elevated the launch beyond typical tech announcements. Instead of leading with internal benchmarks, they showcased enterprise adoption with quantified results. For example, when Bridgewater Associates reports 50-70% faster investment analysis, that carries more weight than any technical score.reuters+1

Competitive Positioning

Competitive Positioning was masterfully subtle. Whilst OpenAI chases consumer popularity and Google talks about broad AI integration, Anthropic positioned themselves as the “serious AI for serious work” company. Additionally, the safety emphasis (ASL-3 protections ) creates differentiation through responsibility rather than pure capability claims.anthropic+3

“In a market saturated with AI hyperbole, Anthropic chose concrete demonstration over abstract speculation—and that’s exactly why it cuts through the noise.”

Where the Marketing Machine Stuttered

Demo Selection Issues

Demo Selection revealed strategic tunnel vision despite the broader portfolio. Although they covered more ground than initially apparent, the examples still skewed heavily technical. Moreover, where were the demos for HR using document templates, marketing teams creating presentation decks, or operations teams managing complex spreadsheets with embedded formulas?

Missing Pricing Conversation

The pricing conversation remained conspicuously absent from demos. At $3/$15 per million tokens, Claude Sonnet 4.5 costs significantly more than alternatives. However, none of the videos addressed cost-benefit analysis or usage optimisation. Therefore, enterprise buyers need ROI justification with real numbers, not just capability proof.cnbc+1

Cross-Demo Integration Problems

Cross-Demo Integration remained unexplored. How do these capabilities work together? For instance, can Claude maintain context from a Catan-style strategic session whilst generating documents and managing browser workflows? Unfortunately, the synergistic potential of combined capabilities never gets addressed in marketing materials.

Geographic Market Blindspot

Geographic Market Considerations were notably absent. The demos and testimonials skewed heavily towards North American enterprises. Furthermore, regulatory considerations, data sovereignty requirements, and regional compliance needs barely featured in the launch narrative.

The UX Writer’s Reality Check: Clear Communication, Missing Context

Purple-themed infographic showing progressive disclosure hierarchy from simple to complex, split between 'Communication Wins' in green featuring plain language, safety messaging, clear structure, and polished demos, versus 'Context Gaps' in orange showing missing error states, hidden complexity, steep learning curves, and enterprise integration challenges

Copy That Actually Connects

Progressive Disclosure Excellence

Progressive Disclosure in the video scripts was genuinely impressive. For example, the Catan demo started with simple game concepts before revealing sophisticated memory management. Similarly, the Chrome extension demo began with familiar workflow problems before showcasing autonomous browser navigation. As a result, viewers could follow the complexity curve without getting lost.

Plain Language Wins

Plain Language Wins throughout all videos. Instead of “enhanced agentic capabilities with improved context window management,” we got “Claude remembers what happened and learns from each game.” Likewise, “real-time software generation” became “Claude builds whatever you imagine on the fly.” This translation from technical specification to human benefit was UX writing excellence.

Safety Communication Success

Safety Communication in the Chrome demo addressed user anxiety head-on. Phrases like “You’re still in control of final edits,” “granular permissions,” and “always asks before sensitive actions” directly tackled the automation fear factor. Moreover, this proactive reassurance demonstrated sophisticated understanding of user mental models.

Document Creation Clarity

Document Creation Clarity was particularly effective: “Claude can now create docs, presentations, and spreadsheets that you can download and edit.” This immediately communicates tangible value without technical jargon. Additionally, the emphasis on “download and edit” addresses the crucial “what do I actually get?” question.reuters

Where Words Failed Users

Failure State Transparency Gap

Failure State Transparency remained largely absent. Whilst the demos showed successful outcomes, they didn’t address what happens when Claude misinterprets instructions, makes errors, or encounters unexpected situations. Therefore, users need to understand not just success scenarios but failure modes and recovery options.

Technical Requirements Buried

Technical Requirements got buried in excitement. How much computational power does real-time software generation require? Similarly, what browser configurations work best with the Chrome extension? Furthermore, how does context management affect performance across different use cases?

Learning Curve Expectations Understated

Learning Curve Expectations were understated across all demos. Although the videos made Claude look immediately useful, they didn’t communicate the iteration and refinement required to maximise effectiveness. Consequently, this sets unrealistic expectations about deployment simplicity.

Enterprise Integration Reality Missing

Enterprise Integration Reality wasn’t adequately addressed. How does Claude work with existing security policies, approval workflows, and compliance requirements? Therefore, the gap between demo simplicity and enterprise complexity needs clearer communication.reuters+1

“Great UX writing acknowledges what doesn’t work as clearly as what does—and these demos consistently stayed in the happy path.”

The AI Evangelist’s Enthusiastic Assessment: Future-Forward, Finally Delivered]

Horizontal timeline infographic showing AI evolution from basic tool to digital colleague, featuring five coloured milestones for context management, real-time generation, browser integration, document creation, and agent SDK, with ASL-3 safety banner below highlighting responsible AI development principles

Revolutionary Capabilities That Actually Revolutionise

Context Management Breakthrough

Context Management Breakthrough represents genuine technical advancement. Watching Claude maintain strategic awareness across 75 minutes of gameplay whilst automatically removing irrelevant information solves a fundamental AI limitation. Moreover, this isn’t incremental improvement—it’s categorical capability expansion.

Real-Time Software Generation Revolution

Real-Time Software Generation delivers on AI promises that have been theoretical for years. Instead of writing code that gets compiled into software, Claude generates functional interfaces directly. Furthermore, this approach could fundamentally change how we think about software development, deployment, and customisation.

Computer Use Integration Success

Computer Use Integration bridges the gap between AI intelligence and practical utility. Rather than requiring custom integrations for each application, Claude works with existing tools through visual interface interaction. Additionally, this universal approach to computer literacy represents a genuine breakthrough in AI-human collaboration.

Document Generation Bridge

Document Generation Integration represents a crucial bridge between AI intelligence and business utility. Rather than requiring users to copy-paste AI output into their tools, Claude generates native files that work seamlessly with existing software ecosystems. As a result, this eliminates friction that has plagued AI adoption in professional environments.reuters

The Agent SDK Revolution

Agent SDK Democratisation potentially transforms the entire AI ecosystem. By open-sourcing the infrastructure that powers their most sophisticated applications, Anthropic enables thousands of developers to build production-ready agents. Furthermore, this platform approach creates network effects that could establish Claude as the de facto standard for agent development.fastmcp+1

The Paradigm Shift in Plain Sight

Systemic Integration Achievement

What excites AI advocates most isn’t individual features—it’s the systemic integration these demos represent. We’re witnessing AI that doesn’t just respond to prompts but maintains strategic awareness, generates solutions dynamically, integrates seamlessly with existing workflows, and produces professional deliverables in standard formats.

Memory and Learning Evolution

The memory and context management shown in the Catan demo suggests AI that builds institutional knowledge rather than starting fresh with each interaction. Similarly, the real-time generation capabilities point toward software that adapts to user needs rather than forcing users to adapt to predetermined functionality.

Browser-Native Integration

Browser-native AI integration demonstrated in the Chrome extension represents the holy grail of seamless AI assistance. Specifically, it’s technology that works where users already work without requiring workflow disruption. Meanwhile, the document creation capability acknowledges that AI success depends on seamless integration with existing business processes.reuters

Safety-First Development

Safety-First Development positions Anthropic as industry leaders in responsible AI scaling. The ASL-3 protections implement sophisticated security measures including egress bandwidth controls, two-party authorisation systems, and enhanced change management protocols. Moreover, this proactive approach to safety could become increasingly important as AI capabilities expand and regulatory scrutiny intensifies.anthropic+1

“We’re witnessing the transition from AI as sophisticated tool to AI as digital colleague—and these demos prove it’s no longer theoretical.”

The End User’s Ground Truth: Impressive Demos, Implementation Questions

Split comparison infographic with 'Demo Promise' in green showing seamless workflows, professional quality, delighted users, and perfect integration against 'User Reality' in red displaying inconsistent results, hidden costs, significant learning curves, variable quality, and compatibility issues, balanced by central scale showing expectations versus reality gap

What Users Actually Need (And Usually Got)

Practical Problem Solving Success

Practical Problem Solving was the demos’ strongest suit. For instance, the home renovation budget scenario directly addressed the universal frustration of scattered information across multiple platforms. Additionally, the Catan game showed AI that learns and adapts over time—exactly what users want from long-term AI assistance.

User Control Emphasis

User Control Emphasis in the Chrome demo addressed the biggest user concern about AI automation: loss of agency. By demonstrating granular permissions and final approval processes, Anthropic showed they understand that empowerment beats automation in user acceptance.

Workflow Integration

Workflow Integration across all demos showed AI working within existing tools rather than requiring new platforms or processes. Furthermore, this approach respects user investment in current systems whilst adding intelligence layers on top.

Document Workflow Integration

Document Workflow Integration was perhaps the demos’ strongest practical appeal. The ability to generate professional-quality Word docs, PowerPoint presentations, and Excel spreadsheets directly addresses the “output formatting” friction that has limited AI adoption in business environments.reuters

Enterprise Social Proof

Enterprise Success Stories provided crucial social proof. When TELUS reports 500,000 hours saved and $90 million in measurable benefit, or when Bridgewater Associates achieves analyst-level precision in investment research, users can visualise concrete applications in their own contexts.datastudios

The Reality Check Gaps

Performance Expectations Mismatch

Performance Expectations weren’t adequately managed. Although the demos showed polished outcomes, they didn’t address typical performance variability, error rates, or the iteration required to achieve shown results. Consequently, this creates expectations gaps that lead to user disappointment.

Cost Transparency Issues

Cost Transparency remained absent from user-facing content. Users need to understand not just what Claude can do, but what sustained usage costs for their specific scenarios. Moreover, the pricing model complexity could surprise users who expect straightforward subscription pricing.techcrunch+1

Learning Investment Requirements

Learning Investment requirements were underplayed. Maximising Claude’s demonstrated capabilities requires prompt engineering skills, workflow optimisation, and ongoing refinement that casual users might not anticipate.

Quality Consistency Questions

Quality Consistency wasn’t demonstrated across the document creation capability. Although the demo showed polished outcomes, users need to understand typical quality variation, common error patterns, and editing requirements for different document types.reuters

Template Limitations

Template Limitations remained unexplored in the file creation demo. Can Claude work with existing company templates? Similarly, how does it handle brand guidelines, formatting standards, and organisational style requirements? These practical constraints significantly impact real-world utility.reuters

“The best AI demos show not just capability but the path to capability—and these videos focused more on destination than journey.”

The Strategic Synthesis: What This Complete Launch Portfolio Signals

Industry Positioning Revolution

Comprehensive Professional AI Strategy

Anthropic’s five-demo strategy plus enterprise validation represents a fundamental shift toward comprehensive professional AI rather than point solutions. By demonstrating capabilities across strategic thinking, creative generation, workflow automation, document production, and enterprise deployment, they positioned Claude as professional AI infrastructure rather than task-specific tool.datastudios+1

Customer-Led Narrative Success

The customer-led narrative particularly signals enterprise-ready AI. Whilst competitors focus on conversational interfaces, Anthropic demonstrated AI that produces measurable business outcomes for recognisable enterprises. Furthermore, this approach acknowledges that professional AI success depends on proven ROI rather than impressive capabilities.datastudios+1

Safety as Competitive Advantage

Safety as Competitive Advantage emerged consistently across the launch. Although competitors emphasise capability expansion, Anthropic positioned responsible AI development as core value proposition. Additionally, the ASL-3 protections could become increasingly important as AI deployment expands and regulatory scrutiny intensifies.anthropic+1

Platform Play Strategy

The Agent SDK strategy represents a sophisticated platform play disguised as developer generosity. By making agent development infrastructure easier to access, Anthropic creates network effects that could establish Claude as the standard foundation for enterprise AI applications.anthropic+1

The Marketing Evolution Reflection

From Promise to Proof

These demos signal the maturation of AI marketing from promise-based to proof-based communication. Early AI companies sold potential; however, Anthropic sold performance backed by enterprise validation. Moreover, this shift suggests the industry moving beyond hype cycle toward utility demonstration.truefan+1

Video-First Communication

Video-First Communication proved more effective than traditional tech marketing approaches. Rather than press releases with embedded demos, Anthropic led with video content that let capabilities speak for themselves. As a result, this approach builds trust through transparency rather than assertion.heygen+1

Customer Success Integration

Customer Success Integration elevates product launches beyond feature announcements toward market validation demonstrations. When major enterprises provide specific metrics about AI impact, marketing transforms from persuasion to proof-of-concept sharing.datastudios

How I Would Have Approached This Launch Differently

From a Digital Marketing Reimagining

Integrated Workflow Narrative

Integrated Workflow Narrative would connect all five demos into cohesive professional AI story. Instead of separate capability demonstrations, I’d show how strategic thinking (Catan) informs document creation, which feeds into browser automation, supported by real-time generation capabilities.

Document Creation Prominence

Document Creation Prominence would be significantly increased. This capability arguably has the broadest professional appeal but received minimal marketing emphasis compared to more technical demos. Furthermore, generating Word docs, PowerPoint presentations, and Excel spreadsheets directly addresses universal business needs.reuters

Cost-Benefit Integration

Cost-Benefit Integration would be central, not peripheral. Every demo would include realistic usage scenarios with associated costs, helping users understand not just what’s possible but what’s practical for their budgets. Additionally, this transparency builds trust whilst managing expectations.cnbc+1

Regional Market Customisation

Regional Market Customisation would acknowledge global enterprise diversity. Different regions have varying regulatory requirements, data sovereignty concerns, and enterprise software preferences that affect Claude adoption strategies.datastudios

From a UX Writing Enhancement

Cross-Demo Integration Guidance

Cross-Demo Integration Guidance would explain how capabilities work together rather than in isolation. Users need clear workflows that show optimal combinations of strategic thinking, document creation, browser automation, and real-time generation.reuters

Failure State Documentation

Failure State Documentation would get equal emphasis with success scenarios. Users need to understand when and why Claude might struggle, how to recognise error states, and what recovery options exist. Moreover, this honesty builds confidence rather than creating unrealistic expectations.

Document Quality Framework

Document Quality Framework would set realistic expectations for file creation capabilities. Different document types (reports, presentations, spreadsheets) require different approaches and produce different quality levels that users should understand upfront.reuters

Technical Requirements Front-Loading

Technical Requirements would be front-loaded, not buried. Each demo would clearly communicate computational needs, optimal configurations, and environmental dependencies that affect performance. Therefore, users can assess feasibility before commitment.

From an AI Advocacy Perspective

Professional AI Maturation Narrative

Professional AI Maturation Narrative would position this launch as industry inflection point. We’re moving from AI assistance to AI collaboration, enabled by comprehensive capability integration and professional workflow compatibility.reuters

Agent SDK as Democratisation

Agent SDK as Democratisation would frame the platform strategy as equalising access to sophisticated AI infrastructure across organisation sizes. Furthermore, this positioning emphasises AI as empowerment tool rather than replacement technology.fastmcp+1

Broader Ecosystem Integration

Broader Ecosystem Integration would position Claude within the expanding AI landscape rather than as isolated capability. This would help users understand how Claude complements existing tools, integrates with current workflows, and fits into organisational AI strategies.

Ethical Implementation Framework

Ethical Implementation Framework would be prominently featured. Although safety got mentioned, deeper discussion of responsible AI deployment, bias considerations, and organisational change management would demonstrate thought leadership beyond technical capability.anthropic+1

From an End User Experience Design

Integrated Workflow Trials

Integrated Workflow Trials would let users experience capability combinations rather than isolated features. This would show how strategic planning informs document creation, which enables browser automation workflows.reuters

Document Template Integration

Document Template Integration would be demonstrated prominently. Users need to see how Claude works with their existing templates, formatting requirements, and organisational standards before committing to adoption.reuters

Implementation Support

Implementation Support would be comprehensive, not superficial. Step-by-step deployment guides, optimisation workshops, and ongoing consultation help users achieve demonstrated value rather than struggling with adoption.

Community Integration

Community Integration would connect users with peers facing similar challenges. Forums, case study sharing, and collaborative optimisation create network effects that enhance individual success.datastudios

Comprehensive Excellence, Integration Opportunity

Technical Achievement vs Accessibility Balance

Anthropic delivered technically impressive demonstrations with strategically comprehensive positioning across five distinct professional AI applications plus enterprise validation. However, they missed opportunities to show how these capabilities integrate into cohesive professional workflows rather than isolated productivity enhancements.datastudios+1

The launch succeeds at establishing comprehensive technical credibility backed by enterprise social proof. Yet it struggles to demonstrate integrated utility that transforms professional work rather than merely assisting with discrete tasks.

Document Creation Undervalued

The document creation capability represents perhaps the most immediately useful feature. However, it received disproportionately limited marketing emphasis compared to more technically sophisticated demos.reuters

Industry Proof vs User Understanding

For an industry desperately trying to prove practical AI value , Anthropic demonstrated comprehensive goods backed by customer validation across the professional workflow spectrum. Nevertheless, they just need marketing that emphasises workflow transformation over isolated capability demonstration.truefan+1

The Real Test

The real test isn’t whether these individual demos impress professionals or whether enterprise customers provide compelling testimonials. Instead, it’s whether users understand how to combine strategic AI thinking, document generation, workflow automation, creative capabilities, and agent development into transformative professional practices that justify significant ongoing investment.

Safety Positioning Advantage

Safety and responsibility positioning provides crucial differentiation in an industry increasingly scrutinised for deployment practices. However, this advantage only matters if users understand how responsible AI development translates into reliable, trustworthy professional tools they can depend on for important work.anthropic+1

Future Marketing Evolution

Ultimately, these videos plus enterprise validation represent the future of professional AI marketing. This includes comprehensive capability coverage, workflow integration focus, professional deliverable emphasis, and customer success validation.

Nevertheless, the next evolution requires demonstrating how these capabilities work together to transform professional work rather than merely augment existing tasks with isolated AI assistance.

Platform Strategy Potential

The Agent SDK strategy potentially creates the ecosystem effects that could make this integration story inevitable rather than aspirational. However, this only works if developers build applications that showcase capability combinations rather than single-purpose solutions.anthropic+1

Internal Links:

Apple’s iPhone 17 Pro Marketing Analysis – examining how tech companies position premium product launches
Samsung India Digital Marketing Campaign Analysis – comparing enterprise vs consumer positioning strategies
September Smartphone Marketing Analysis – understanding launch timing and competitive positioning

Footnotes:

Anthropic, “Introducing Claude Sonnet 4.5,” September 28, 2025anthropic
TechCrunch, “Anthropic launches Claude Sonnet 4.5,” September 29, 2025youtube
CNBC, “Anthropic launches Claude Sonnet 4.5,” September 29, 2025techxplore
AWS, “Introducing Claude Sonnet 4.5 in Amazon Bedrock,” September 28, 2025techcrunch
Reuters, “Anthropic launches Claude 4.5,” September 29, 2025anthropic
Anthropic, “Create polished files in minutes with Sonnet 4.5,” September 29, 2025reuters
Claude Plays Catan Demo Analysis DocumentClaude-Plays-Catan__-Context-Management-Demo-Brea.docx
Imagine with Claude Real-Time Software Generation Analysis DocumentImagine-with-Claude__-Real-Time-Software-Generati.docx
Claude for Chrome Product Demo Analysis DocumentClaude-for-Chrome__-Product-Demo-Breakdown.docx
TrueFan AI, “AI Video Creation Guide,” September 15, 2025truefan
HeyGen, “Video Marketing Strategies Using AI Tools,” December 12, 2024heygen
FastMCP, “Anthropic Launches Claude Agent SDK,” September 28, 2025fastmcp
DataStudios, “Claude in the enterprise: case studies,” August 30, 2025datastudios
Anthropic, “Activating AI Safety Level 3 Protections,” May 21, 2025anthropic
Anthropic, “Activating AI Safety Level 3 protections,” May 21, 2025anthropic
JetBrains, “Introducing Claude Agent in JetBrains IDEs,” September 29, 2025jetbrains
Anthropic, “Building agents with the Claude Agent SDK,” September 28, 2025anthropic
Economic Times, “Anthropic launches Claude 4.5,” September 28, 2025economictimes