GPT-5.4, $100M in Claude Partners, and the Court Ruling That Changes Agentic AI

Executive Summary

OpenAI shipped GPT-5.4 on March 5, adding native computer-use capabilities and a one-million-token context window to the GPT-5 family. Anthropic launched its Claude Partner Network with a $100 million commitment, formalizing enterprise channels through Accenture, Deloitte, Cognizant, and Infosys. Perplexity had a consequential week on two fronts: it unveiled "Personal Computer," an always-on AI agent running on dedicated Mac hardware, while a federal judge issued a temporary injunction blocking its Comet browser from accessing Amazon's marketplace. These three stories together reflect where enterprise AI is heading — capable agentic systems, structured partner ecosystems, and intensifying legal scrutiny around what AI agents are actually permitted to do.

Top Stories

1. OpenAI Releases GPT-5.4 with Native Computer Use

OpenAI GPT-5.4

OpenAI released GPT-5.4 and GPT-5.4 Pro on March 5, describing it as its most capable model for professional work. The release introduced several meaningful upgrades over GPT-5.2: native computer-use mode (the first general-purpose OpenAI model capable of autonomously working across applications), a one-million-token context window via API, 33% fewer factual errors, and a tool-search mechanism that reduces token costs by 47% in tool-heavy workflows without accuracy loss. GPT-5.4 also launched with financial plugins for Microsoft Excel and Google Sheets, targeting enterprise document and data workflows directly.

GPT-5.4 Thinking is available to Plus, Team, and Pro subscribers. GPT-5.4 Pro is available to Pro and Enterprise plan holders. The prior GPT-5.1 model family was deprecated from ChatGPT effective March 11, with users automatically migrated to corresponding GPT-5.3/5.4 variants.

Business Impact: The native computer-use capability marks a meaningful step for enterprises building agentic workflows. GPT-5.4 can now operate across desktop applications autonomously, which opens practical deployment paths for back-office automation, QA, and software testing tasks that previously required custom orchestration layers.

2. Anthropic Launches $100M Claude Partner Network

Anthropic Claude Partner Network

Anthropic officially launched the Claude Partner Network on March 12, committing $100 million for 2026 to support organizations that help enterprises deploy Claude. The network formalizes a partner ecosystem that includes Accenture (training 30,000 professionals on Claude), Cognizant (opening Claude access across its 350,000-person global workforce), Deloitte, and Infosys. Membership is free for any firm bringing Claude to market.

The program includes access to Anthropic Academy training materials, dedicated Applied AI engineers for live customer deals, sales playbooks, co-marketing support, and a new "Claude Certified Architect, Foundations" technical certification. Anthropic is scaling its partner-facing headcount fivefold and adding localized go-to-market support across international markets. A Code Modernization starter kit targeting legacy codebase migration was released alongside the network launch.

Business Impact: This is a direct signal that Anthropic is moving from a model provider to a full enterprise software partner. The scale of partner enablement, combined with the Code Modernization toolkit, positions Claude for large-scale engineering projects rather than just conversational interfaces.

3. Perplexity Launches "Personal Computer," Then Gets Blocked on Amazon

Perplexity Comet browser

Perplexity had two significant, contrasting developments this week. At its Ask 2026 developer conference, the company unveiled Personal Computer — software that runs continuously on a Mac mini to give Perplexity's cloud AI agent persistent access to local files, applications, and sessions. It also launched Computer for Enterprise, which connects to corporate tools including Snowflake, Salesforce, and HubSpot through app connectors, with SOC 2 Type II compliance, SAML SSO, and sandboxed query execution. Personal Computer is priced at $200/month for Max subscribers and requires a waitlist at launch. New developer APIs for Search, Agent, Embeddings, and Sandbox were also released. PitchBook announced a partnership on March 12 to surface firmographic data directly inside Perplexity's conversational interface.

Simultaneously, on March 10, U.S. District Judge Maxine Chesney granted Amazon a temporary injunction barring Perplexity's Comet browser from accessing Amazon's systems. The court found that while Comet accessed accounts with user permission, that permission did not constitute authorization from Amazon itself. Perplexity must destroy all Amazon data collected through Comet and has until March 16 to seek a stay from the Ninth Circuit.

Business Impact: Personal Computer and Computer for Enterprise represent a serious enterprise product push from Perplexity. The Comet injunction, however, sets an important legal precedent: user consent alone may not be sufficient for AI agents to operate inside third-party platforms. Any enterprise deploying agentic AI that touches external systems needs to examine the authorization framework closely.

4. Agentic AI Startups Continue Attracting Major Investment

Agentic AI startups

Investment in AI agent infrastructure continued at pace this week. Israeli customer service AI startup Wonderful raised $150 million in a Series B at a $2 billion valuation, led by Insight Partners, with Index Ventures, IVP, and Bessemer participating. Lyzr AI, which builds enterprise agent infrastructure, closed a round led by Accenture that quintupled its valuation to $250 million. Dify, which offers a no-code visual builder for multi-agent applications, raised $30 million at a $180 million valuation.

The pattern across these rounds reflects a consistent investment thesis: vertical AI agents for specific industries (telecom, finance, manufacturing, healthcare) with measurable automation outcomes are commanding premium valuations at early stages.

Business Impact: For enterprises evaluating AI vendors, these rounds signal that the agentic AI market is consolidating quickly around infrastructure players. Organizations that wait to standardize their agentic tooling risk building on platforms that may be displaced within 18–24 months.

Quick Bytes

  • EU AI Act Code of Practice: The European Commission published a second draft of the Code of Practice on Marking and Labelling of AI-generated content on March 5. Full transparency and high-risk system requirements under the AI Act come into force August 2, 2026. (Source: EU Commission, March 5)
  • Washington chatbot safety bill: Washington state lawmakers passed HB 2225, a companion chatbot safety bill, making it the second U.S. state in 2026 (after Oregon) to enact specific AI companion safety legislation. (Source: State Legislature, March 2026)

Industry Impact Analysis

GPT-5.4's computer-use capabilities and Perplexity's Personal Computer both reflect the same underlying shift: AI is moving from text generation to autonomous task execution across real systems. The legal context matters here. The Amazon vs. Perplexity ruling establishes that user authorization and platform authorization are legally distinct. As enterprises deploy agents that interact with third-party APIs, databases, and SaaS platforms, they need to confirm that terms of service and data agreements cover agentic access, not just human access.

Anthropic's partner network investment signals that model providers are entering the professional services space directly. For software development partners and system integrators, this changes the competitive dynamic. Being certified on Claude — and part of that delivery network — is increasingly a differentiator.

The ongoing investment surge in agentic infrastructure confirms that the automation layer sitting between AI models and enterprise applications is where capital is concentrating. The practical question for product and engineering leaders is no longer "should we use AI agents" but "which infrastructure are we standardizing on, and why."

About Azumo

Azumo helps companies design, build, and scale AI-powered software using the latest models and agentic frameworks. From enterprise integrations to custom AI agents, the team brings technical depth and delivery speed to complex AI projects.

If you are evaluating agentic AI implementations, Azumo brings deep implementation expertise across the leading model platforms, including Claude, GPT-5.4, and multi-agent orchestration.

Sources

This newsletter is curated by Azumo's AI Intelligence Scanner to help enterprises stay ahead of critical AI developments that impact business strategy and technology roadmaps.

Are You New to Outsourcing?
We Wrote the Handbook.

We believe an educated partner is the best partner. That's why we created a comprehensive, free Project Outsourcing Handbook that walks you through everything from basic definitions to advanced strategies for success. Before you even think about hiring, we invite you to explore our guide to make the most informed decision possible.