Salesforce has launched Agentforce 360 — a global, enterprise-grade platform to build, deploy and govern AI agents across CRM/ERP workflows — and simultaneously expanded deep model partnerships (OpenAI GPT‑5, Anthropic Claude and others), new developer tooling (Agentforce Vibes / Vibe Codey), voice-enabled agents, Agent Script (a portable agent scripting language), Agentforce Commerce (ChatGPT Instant‑Checkout integration) and a Slack-native AgentExchange marketplace; the company says Agentforce 360 is generally available as of Oct 13–14, 2025. (investor.salesforce.com)

This move reframes Salesforce from a traditional CRM/vendor into an orchestration layer for ‘agentic enterprises’ — combining data (Data 360/Tableau), governance, multiple LLM choices, telephony/voice, and commerce flows so companies can automate customer/service/sales workflows while keeping data controls; it also ties Salesforce tightly into the broader AI/LLM ecosystem (OpenAI, Anthropic, Google Cloud, Amazon Bedrock) which could accelerate enterprise AI adoption but raises questions about cost, governance, and vendor lock‑in. (investor.salesforce.com)

Primary players are Salesforce (product owner / orchestrator), OpenAI (GPT‑5 / ChatGPT integrations), Anthropic (Claude model availability), Slack (AgentExchange marketplace / Slack integrations), Tableau/Data 360 (analytics & governance within the stack), major contact‑center partners (Amazon Connect/Bedrock, Five9/Genesys/Avaya/Vonage integrations) and a broad partner ecosystem (Google Cloud, Amazon, Perplexity, Vercel, WRITER and many ISV/partners); investors, enterprise customers (including public sector pilots) and competitors (ServiceNow, contact‑center vendors) are active stakeholders. (investor.salesforce.com)

At Oracle AI World (Oct 13–16, 2025) Oracle crystallized a practical, enterprise-first AI strategy: it launched the Oracle AI Factory and pushed agent-first capabilities (AI Agent Studio / Fusion agents and an Agent Marketplace) plus the Autonomous AI Lakehouse to embed AI where enterprise data lives, and announced multicloud and hyperscaler partnerships and hardware deals (including a publicized AMD MI450 GPU deployment plan) to scale inference/training capacity for ERP/CRM workloads and partner ecosystems. (oracle.com)

This matters because Oracle is shifting AI messaging from hypothetical demos to production mechanics for ERP/CRM: embedded role‑specific agents in Fusion apps, data‑first governance (data sovereignty/multicloud) and partner-led last‑mile extensions aim to accelerate measurable ROI (reduced cycle times, lower abandonment, automated ledger/HR/scm tasks) while creating a partner marketplace and SI opportunity — all of which changes how enterprises plan, buy and operate AI in core systems and raises new governance, procurement and vendor‑competition questions. (oracle.com)

Primary actors are Oracle (executives and product leads including Steve Miranda and co‑CEO Clay Magouyrk / Larry Ellison as strategy figureheads) as the platform and infra provider; hyperscalers and hardware vendors (AMD, plus reported hyperscaler collaborations) for compute and multicloud reach; system integrators/partners (Mythics, PwC, DXC, many Oracle partners) and early adopter customers (Marriott, Grupo Bimbo, Etihad Salam and others highlighted in customer stories) who validate enterprise use cases. OpenAI and other AI innovators/startups appear in the ecosystem as large customers/partners for OCI infrastructure. (oracle.com)

Oracle announced on Oct 14, 2025 two complementary infrastructure moves: (1) it will be a launch partner to deploy an initial 50,000 AMD Instinct MI450 GPUs as a publicly available AMD-powered AI supercluster starting in calendar Q3 2026 (with further expansion into 2027+), and (2) it unveiled OCI Zettascale10 — a next‑generation, NVIDIA‑powered zettascale fabric that Oracle says can connect hundreds of thousands of NVIDIA GPUs (targeting up to 800,000 GPUs) and deliver up to 16 zettaFLOPS peak — reflecting a dual‑vendor strategy for large‑scale AI compute. (oracle.com)

This matters because Oracle is explicitly diversifying cloud AI compute away from single‑vendor dependence on NVIDIA by offering both AMD rack-scale MI450 capacity (improves supply‑chain resilience and price/choice for customers) and continuing to build enormous NVIDIA zettascale clusters (Zettascale10) optimized for very large training clusters — a move that affects hyperscaler competition, chip vendor market share, enterprise access to large model training/inference, and Oracle’s ability to embed and scale generative‑AI features across its ERP/CRM application portfolio. The announcements also tie into Oracle’s broader product push (AI Agent Studio, Fusion AI agents and marketplace) to embed LLM/agent capabilities into ERP/CRM workflows. (reuters.com)

Primary players are Oracle (OCI, Zettascale10, AI Agent Studio/Fusion Applications), AMD (Instinct MI450 GPUs, Helios/Helios rack design), NVIDIA (providing the GPUs and stack for Zettascale10), and large AI customers/partners such as OpenAI (Stargate/Abilene supercluster collaboration). Oracle executives publicly named include Mahesh Thiagarajan (OCI EVP) and Karan Batta (SVP, OCI); AMD spokespeople/executives (e.g., Forrest Norrod/AMD leadership) also featured in coverage. These companies and people are coordinating hardware, networking (Oracle Acceleron RoCE), software stacks, and application embedding for ERP/CRM use cases. (oracle.com)

Oracle announced "Oracle AI Database 26ai" (marketed as a long-term-support release replacing 23ai) on October 14, 2025 — a major database release that embeds AI across the core data stack (AI Vector Search, agentic AI support, Model Context Protocol (MCP), Apache Iceberg support via an Autonomous AI Lakehouse, built‑in privacy controls, Exadata acceleration, and quantum‑resistant encryption), and Oracle published both product/installation documentation and Free RPMs for Oracle Linux 9 so developers can install a "Free" edition locally or in cloud environments. (oracle.com)

This matters because Oracle is moving from 'AI on top of data' to 'AI architected into the database' — enabling ERP/CRM and other operational systems to run agentic AI workflows, vector/semantic retrieval, and RAG-style queries directly against private enterprise data without copying it to external LLM services; that has direct implications for faster AI-enabled business processes (e.g., automated case handling, intelligent recommendations, assisted order-to-cash workflows), tighter data governance/compliance, multicloud deployment flexibility, and potential cost/performance differences when combined with Exadata and protected private AI containers. (oracle.com)

Primary players are Oracle (product, Exadata, OCI and enterprise sales/engineering), hyperscaler partners where Autonomous AI Lakehouse is available (AWS, Microsoft Azure, Google Cloud), AI/LLM model and GPU partners (LLM providers and NVIDIA integrations), enterprise customers running ERP/CRM/workloads, and ecosystem contributors (tooling/agentic frameworks and standards such as Apache Iceberg and MCP). Key Oracle spokespeople quoted in materials include Juan Loaiza (EVP, Oracle Database Technologies) and named analysts (e.g., Holger Mueller). (oracle.com)

Oracle has rapidly expanded its "agentic AI" capabilities inside Oracle Fusion Cloud by shipping AI Agent Studio (a no/low-code environment for building agents), embedding dozens of prebuilt agents across ERP/HCM/SCM/CX, and — as of mid‑October 2025 — launching an AI Agent Marketplace that lets partners publish validated, deployable agents directly into Fusion workflows; IBM announced three partner agents (Intercompany, Smart Sales Order Entry, Requisition‑to‑Contract) available on that Marketplace on Oct 16, 2025. (oracle.com)

This matters because it moves enterprises from experimenting with LLMs toward productionized, workflow‑native automation: prebuilt agents + a partner marketplace shorten time‑to‑value, broaden model choice (Oracle is supporting multiple LLM providers and partner models), and shift the competitive battleground for ERP/CRM/HCM to an ecosystem play — with implications for automation of finance/HR/supply‑chain tasks, vendor lock‑in, governance, and cloud consumption (OCI). (oracle.com)

Primary players are Oracle (owner/operator of Fusion Cloud, AI Agent Studio and the new AI Agent Marketplace), global systems integrators and consultancies (IBM Consulting — which published validated agents — plus Accenture, Deloitte, KPMG, Wipro and others publishing marketplace agents), LLM/model providers (OpenAI, Anthropic, Google, Meta, Cohere, xAI are referenced in partner/coverage summaries), and infrastructure partners (Red Hat/OpenShift for watsonx integration). (oracle.com)

At Dreamforce (Oct 13–16, 2025) Salesforce launched Agentforce 360 — a unified “agentic” AI stack for building, deploying and governing enterprise AI agents — and announced expanded partnerships that embed OpenAI’s frontier models (including GPT-5 via ChatGPT) and Anthropic’s Claude into the Agentforce ecosystem, surface Agentforce apps inside ChatGPT (Apps in ChatGPT) and Slack, and add Agentforce Commerce/Instant Checkout commerce flows; Salesforce positions Agentforce 360 as broadly available with partner model choice (OpenAI, Anthropic, Google Gemini, Amazon Bedrock) and Slack-native agent distribution. (salesforce.com)

This matters because a major CRM/ERP vendor (Salesforce) is turning generative AI models into first‑class components of enterprise workflows and commerce channels — enabling CRM data and Tableau analytics to be queried and acted on from ChatGPT and Slack, expanding model choice and hosting options for regulated industries, and accelerating the shift from point AI pilots to large-scale, agent-driven operational automation; the announcements also triggered investor scrutiny and competing vendor activity as the market recalibrates around data governance, model selection and monetizable distribution (commerce) opportunities. (reuters.com)

Key players are Salesforce (Agentforce 360, Marc Benioff) as the platform and integrator; OpenAI (ChatGPT, GPT‑5) for frontline model and Apps in ChatGPT distribution; Anthropic (Claude, Amazon Bedrock-hosted Claude) for regulated-industry model choices; Slack as the ‘agentic OS’/distribution layer; Google (Gemini) and Amazon (Bedrock) as additional model/hosting partners; customers named in announcements include Reddit, CrowdStrike, RBC Wealth Management, Adecco and enterprise systems like Tableau; financial community and partners (Stripe for Instant Checkout) are also implicated. (salesforce.com)

At Dreamforce 2025 (mid‑October 2025), Salesforce formally launched Agentforce 360 — a unified, enterprise-grade 'agentic' AI stack that combines a governed data layer (Data 360), a new Agentforce platform (Agent Builder, conversational builders, hybrid reasoning, voice), and deep Slack integration positioned as the workplace 'Agentic OS' (Slack‑first apps, a reimagined Slackbot, Channel Expert Agent and a Slack‑native AgentExchange); Salesforce also announced expanded model partnerships (OpenAI, Anthropic, Google) and new industry packs while describing a Customer Zero program to bake real customer learnings into product direction. (salesforce.com)

This matters because Salesforce is using its CRM/ERP‑adjacent platform footprint to push AI agents into core enterprise workflows — shifting CRM/ERP value from static records to autonomous, governed actions across systems (sales, service, commerce, field service, billing) — which could accelerate AI adoption in regulated and mission‑critical environments, alter CIO budgets (toward AI/automation), and intensify competition with incumbents (ServiceNow, ERP vendors) while raising immediate questions about adoption tempo, data governance, and real world ROI. (salesforce.com)

Primary players are Salesforce (Marc Benioff and product teams driving Agentforce 360 and Data 360), Slack (positioned as the 'Agentic OS' — Slack leadership and product execs promoting Slackbot and Slack‑first apps), model partners such as OpenAI and Anthropic (deep integration announced), large cloud and systems partners (Google Cloud, AWS/Amazon Bedrock), enterprise SI and consulting partners (Accenture, Deloitte, IBM, PwC, etc.), and early customer/pilot participants (e.g., Simplyhealth, Barcelona City Council) who serve as 'Customer Zero' inputs for product direction. (salesforce.com)

ServiceNow has launched "AI Experience" (AIx) — a multimodal, agentic AI user interface that embeds role-aware AI agents (AI Lens GA; voice/web agents, AI Data Explorer and an AI-powered CPQ slated by end-2025) across the Now Platform to turn workflows and CRM into an "AI-first" system of action; separately, on July 31, 2025 Salesforce and ServiceNow each committed ~$750M (a combined $1.5B) to Genesys Cloud, a move that deepens partnership ties and values Genesys at roughly $15B while bolstering CX/CCaaS orchestration as enterprises adopt agentic AI. (servicenow.com)

This matters because ServiceNow is attempting to make AI the new enterprise UI — moving from passive records to agents that act — which directly touches ERP/CRM and contact-center stacks; the Genesys investment by two competing enterprise-software vendors signals strategic consolidation around an independent CX orchestration layer (accelerating integrations between CRM, ITSM and CCaaS) while raising questions about vendor influence, data integration, governance, and competitive dynamics between ServiceNow and Salesforce. (cio.com)

Primary corporate players are ServiceNow (CEO Bill McDermott; EVP & Chief Experience Officer Amy Lokey; Amit Zavery cited on product/partnership strategy), Genesys (chairman & CEO Tony Bates), and Salesforce (David Schmaier quoted on partnership); private-equity holders Permira and Hellman & Friedman remain majority owners of Genesys and are part of the capital/repurchase structure. (servicenow.com)

The European Commission opened a formal antitrust investigation into SAP on September 25, 2025, probing whether SAP’s policies for maintenance and support of its on‑premises ERP software distort competition in the aftermarket and limit customers’ ability to use third‑party support — an inquiry that could lead to remedies or fines (reportedly up to ~10% of annual global sales). At the same time, SAP is pushing its AI/ERP agenda (new data/AI products and the SAP Connect forum) and large customers and partners are racing to migrate to S/4HANA/RISE offerings (notably IBM’s internal migration affecting ~150,000 users and claiming ~30% infrastructure cost reduction), all of which frames the probe as a test of how legacy maintenance economics, cloud migration incentives and early enterprise AI rollouts interact. (reuters.com)

This matters because (a) the probe targets SAP’s aftermarket economics — a core revenue stream that funds R&D and cloud transition investments — and could force changes to licensing/maintenance contracting across thousands of enterprise customers in the EEA; (b) any remedy or fine would reshape competitive dynamics between SAP, hyperscalers, and third‑party support providers and could accelerate customers’ moves to cloud/RISE or to third‑party support alternatives; and (c) the regulatory action coincides with SAP’s public push for AI-enabled ERP (SAP Connect coverage and early 'Ask my Payslip' rollouts), meaning regulators will be weighing incumbent market power while customers adopt AI features that depend on data access and support models. (ft.com)

Primary actors are SAP SE (the defendant, pushing S/4HANA, RISE and AI initiatives), the European Commission (competition enforcement; Vice‑President and Commission officials leading the probe), large customers and partners such as IBM (early large S/4HANA/RISE adopter and migration reference), early adopters cited in coverage (PostNL, Heartland Dental), independent analysts and media (diginomica reporting critical and practical takes), and third‑party/support vendors and litigants (e.g., Celonis and other aftermarket providers who have previously raised competition concerns). (reuters.com)

Enterprise CRM/ERP AI adoption is accelerating while governance, data‑sovereignty and trust gaps are becoming front‑and‑centre: vendors (notably Oracle and Salesforce) are pushing integrated AI Data Platforms and agent frameworks that let models reason over sensitive CRM/ERP data without wholesale data movement, while regulators and consumer bodies are flagging risks (and the EU has opened a formal antitrust probe into SAP’s ERP maintenance/support practices). (oracle.com)

This matters because CRM/ERP systems hold mission‑critical, highly regulated and often cross‑border customer and supply‑chain data — so the technical trend of “bringing models to the data” intersects with geopolitics (sovereign compute, localization), competition policy (vendor lock‑in and aftermarket control) and consumer trust/regulatory pressure; empirical analyses and industry disclosures show major infrastructure and investment shifts (local/cloud partnerships, sovereign compute builds) that change who controls access, compliance and liability for enterprise AI. (arxiv.org)

Primary players are large enterprise-software vendors (Oracle, SAP, Salesforce) pushing AI features into Fusion/ERP/CRM stacks; hyperscalers and infrastructure partners (AWS, Microsoft Azure, Google Cloud, AMD/NVIDIA) supplying sovereign compute; regulators and enforcers (European Commission) pursuing competition and consumer-protection actions; and civil society/consumer groups (Consumers International) and standards bodies driving trust/transparency demands. (reuters.com)

Cloud vendors, CRM/ERP platforms, and integration toolmakers are converging on richer, bidirectional connectors and agent protocols so AI agents can both READ from and ACT on enterprise systems (CRM/ERP) — examples include AWS Glue adding write operations to SAP OData, Marketo, Salesforce Marketing Cloud and HubSpot (Oct 3, 2025), multi‑cloud patterns for moving ERP/Fusion data into Azure Fabric, and the rise of MCP (Model Context Protocol) wrappers that let Oracle/OCI and other cloud services expose CRM/ERP functionality to LLM agents (Dev.to guides and AWS examples). (aws.amazon.com)

This matters because enterprises need safe, auditable ways for AI agents to act (create/update records, trigger workflows) across CRM and ERP without brittle point-to-point scripts — the combination of native write-capable connectors (ETL/Glue), standardized agent-to-tool protocols (MCP), and platform integrations (Salesforce Agentforce <> Amazon Bedrock Agents, HubSpot agent builders) enables end-to-end agentic automation but also raises governance, identity, and data-trust requirements that change how IT, security and compliance operate. (aws.amazon.com)

Key players span cloud providers (AWS, Microsoft/Azure, Oracle/OCI, Google Cloud), CRM/ERP vendors (Salesforce, HubSpot, SAP, Oracle Fusion), AI-model and protocol providers (OpenAI, Anthropic/MCP, Amazon Bedrock), integration/automation vendors (Workato, iPaaS players), and developer communities producing how‑tos (DEV Community, AWS blogs). Strategic partnerships (Salesforce + OpenAI/Anthropic + Bedrock; Anthropic/MCP adoption) and vendor features (Glue connectors, Agentforce integrations) are driving the ecosystem. (salesforce.com)

At Workday Rising 2025 (announced Sept 16, 2025) Workday laid out an AI-first product and platform roadmap for ERP/CRM-adjacent work: it announced a definitive agreement to acquire AI knowledge-and-learning firm Sana for about $1.1 billion, launched Workday Data Cloud (zero-copy, partner integrations with Databricks, Salesforce Data Cloud and Snowflake), expanded its Illuminate agent portfolio and Agent System of Record (ASOR), and announced agent interoperability with Microsoft (registering agents built on Azure AI Foundry/Copilot Studio/Entra Agent ID into Workday's ASOR). The company also highlighted new developer/build capabilities (a Build/Extend story) and timing for early access / GA of agent and data features. (investor.workday.com)

This matters because Workday is positioning itself as a ‘front door for work’ that unifies HR and finance context with CRM and operational data to power agentic AI across the enterprise: zero-copy data flows and the ASOR aim to enable real‑time, governed AI actions across multiple vendor stacks, shifting how organizations extract value from ERP/CRM datasets, how partners integrate, how SaaS vendors monetize AI usage (consumption/pricing implications), and how customers think about governance, compliance, and skills/reskilling. The announcements also triggered investor and market responses (activist Elliott stake, expanded buyback/capital return actions), underlining commercial as well as technical stakes. (en-gb.newsroom.workday.com)

Workday (product leadership including Gerrit Kazmaier, Carl Eschenbach, Shane Luke), acquisition target Sana (founder/CEO Joel Hellermark), platform partners Databricks, Salesforce (Data Cloud), Snowflake, cloud/AI partners including Microsoft (Entra Agent ID / Azure AI Foundry / Copilot Studio), AWS/Bedrock and Agent Partner Network firms (Paradox, Glean, others), and investors/market actors such as Elliott Management. Analysts and commentators (e.g., diginomica, WSJ, industry analysts like Josh Bersin) are shaping the narrative and critique. (investor.workday.com)

Two related trends have crystallized in autumn 2025: researchers disclosed a critical prompt‑injection chain (ForcedLeak) in Salesforce's AI agent platform Agentforce that could be used to exfiltrate CRM data via Web‑to‑Lead forms (discovered by Noma, CVSS 9.4; Salesforce patched and enforced Trusted URLs), and security teams / threat intelligence firms (Mandiant/Google Threat Intelligence, CrowdStrike, others) documented active, large‑scale exploitation of zero‑day vulnerabilities in Oracle E‑Business Suite (notably CVE‑2025‑61882) used in extortion campaigns that began in mid‑2025 and triggered emergency patches in early October 2025. (noma.security)

These incidents show two complementary risks to ERP/CRM ecosystems: (1) AI agent integrations expand the attack surface for CRM platforms — enabling indirect prompt injection and automated exfiltration through trusted pipelines — and (2) traditional ERP platforms remain lucrative targets for massized zero‑day exploitation and data‑theft extortion, with unauthenticated RCEs in Oracle EBS leading to wide targeting, data exfiltration and executive extortion demands; together they raise operational, regulatory and supply‑chain risks for enterprises that mix AI agents, third‑party integrations and legacy on‑prem ERP. (noma.security)

Key actors include vendor and research teams (Salesforce and Noma Security for the ForcedLeak disclosure and patch), enterprise defenders and incident responders (Mandiant/Google Threat Intelligence, CrowdStrike), the threat actors leveraging exploits and extortion (actors using the CL0P brand and related groups), regulators/incident‑response bodies (CISA/FINRA/industry ISACs advising rapid patching), and third‑party/managed service ecosystems that connect ERP/CRM systems to clouds and AI stacks. (infosecurity-magazine.com)

Incumbent CRM/ERP vendors and their ecosystems are accelerating a two-track strategy: (1) large platform plays that stitch data, agentic AI and commerce together (Salesforce’s Agentforce 360 launch and expanded partnerships with OpenAI/Anthropic; a pending major Informatica deal) and (2) ecosystem consolidation and verticalization by partners and specialists (Oracle partners and SIs extending AI into Fusion/SCM/ERP, and systems integrators such as Mythics buying SmartERP and SpearMC), while AI‑native startups (for example Aurasell) raise large seed rounds to build alternative, agentic-native CRM experiences. (salesforce.com)

This matters because CRM and ERP are moving from model/feature add-ons (’AI in the UI’) to platform-level agentic architectures that require unified data plumbing, governance, and partner ecosystems — a shift that raises strategic risks and opportunities: incumbent platforms can lock in customers with integrated agent/data stacks and commerce hooks, SI consolidation speeds deployment at scale, and startups target niches or attempt to displace legacy stacks; the result influences competitive dynamics, M&A valuations, and enterprise procurement for years. (salesforce.com)

Major vendors (Salesforce — Marc Benioff and Agentforce 360; Oracle — expanding AI agents & partner/network activity), ERP/CRM incumbents (Workday), systems integrators/partners (Mythics; Intellinum), startups (Aurasell — founders Jason Eubanks & Srinivas Bandi), investors (Next47, Menlo Ventures, Unusual Ventures), and model providers/partners (OpenAI, Anthropic) are the central participants driving product launches, acquisitions and funding. (salesforce.com)

Large cloud vendors, system integrators and ERP vendors are accelerating multicloud SAP/ERP migration and integration strategies that combine cloud-native data fabrics, AI agents and cross‑cloud interoperability: Google Cloud announced SAP Business Data Cloud on Google Cloud and new agent/ADK capabilities on Sep 16, 2025 to push bi‑directional BigQuery–SAP data flows and Vertex/agent-led automation; practitioners are publishing practical patterns for moving Oracle Fusion data into Microsoft Fabric/Azure (example guide published Sep 20, 2025); and major enterprises and integrators (notably IBM with a large RISE with SAP migration) are moving core ERP workloads to cloud platforms (IBM reported moving >150,000 users to SAP S/4HANA with ~30% infrastructure cost reduction), while Oracle’s expanding multicloud and AI infrastructure deals (including very large OpenAI commitments) reshape where ERP/CRM data and AI run and raise new questions about data residency and provider roles. (cloud.google.com)

This matters because ERP/CRM modernization is now tightly coupled to AI and data strategy: platform choices (Google BigQuery + SAP Business Data Cloud, Azure Fabric integrations, IBM Power Virtual Server for RISE, Oracle’s OCI AI scale) determine latency, governance, TCO and how quickly organizations can operationalize generative/agentic AI across finance, supply chain and CRM—so migrations are being driven not just by cost (IBM reported ~30% lower infra costs) but by the need for unified data fabrics, vendor interoperability, and SI ecosystems that can deliver AI-enabled business processes at scale. (news.sap.com)

Primary players are SAP (platform + Joule/Business Data Cloud), hyperscalers (Google Cloud, Microsoft Azure, Oracle Cloud), integrators and SI partners (IBM Consulting, SNP and others), large AI buyers/partners (OpenAI), and specialized tooling/partner authors (community guides for Azure Fabric migrations). These actors are collaborating and competing—Google and SAP deepening integration for BigQuery/agents; Microsoft/Azure and Oracle enabling cross‑cloud data movements and Fabric use cases; IBM packaging RISE on Power Virtual Server and using IBM Consulting as both vendor and early customer. (cloud.google.com)

Enterprise software vendors (Oracle, Workday, Salesforce and others) are embedding purpose-built, agentic and generative-AI capabilities directly into ERP/CRM/HCM suites so that role-specific AI agents automate workflows and surface predictive insights for finance, HR and healthcare use cases — e.g., Oracle announced new Finance and HCM AI agents in October–September 2025 to automate planning, payments, recruiting, career development and payroll analysis, and Workday unveiled Data Cloud, agent builders and HR/Finance agents at Workday Rising in mid‑September 2025, while Salesforce has launched Agentforce 360 and expanded model partnerships to enable healthcare and regulated use cases. (prnewswire.com)

This shift matters because embedding AI inside the system of record (ERP/CRM/HCM) promises faster cycle times (continuous planning, faster close, automated payroll checks), tighter compliance and provenance (agents operating in governed applications), and new vertical workflows for healthcare and finance — potentially reducing headcount on routine tasks, improving forecast accuracy and unlocking industry-specific AI services at scale; it also raises immediate governance, data residency and safety questions for regulated domains. (nasdaq.com)

Major platform vendors and ecosystem partners are leading the move: Oracle (Fusion Cloud + Oracle AI Agent Studio; HR agents announced Sept 16, 2025 and Finance agents Oct 15, 2025), Workday (Workday Data Cloud, Agent builder and Extend/Build announced Sep 16, 2025), Salesforce (Agentforce/Agentforce 360 and integrations with OpenAI/Anthropic announced Oct 13–14, 2025) — alongside model providers (OpenAI, Anthropic), cloud/infra partners and customers (including healthcare providers cited using Salesforce Agentforce and higher‑education customers such as Wellesley/Workday case studies). (oracle.com)

Enterprise AI agents and cross‑platform agent integrations are accelerating into ERP/CRM stacks: major vendors and partners are publishing prebuilt agents to marketplaces (e.g., IBM shipped three Oracle‑validated agents for Oracle Fusion on Oct 16, 2025), cloud providers are publishing reference patterns to connect agent platforms (AWS published a detailed Agentforce ↔ Amazon Bedrock Agents integration guide on Aug 7, 2025), and both vendor ecosystems and startups (OpenAI’s Agent Builder + MCP ecosystem, HubSpot MCP tooling, and new entrants like Aurasell) are rapidly building third‑party connectors and commercial agent products to automate sales, service, HR and supply‑chain workflows. (newsroom.ibm.com)

This matters because agent marketplaces, MCP connectors and cross‑platform orchestration let organizations stitch LLMs/agents into mission‑critical ERP/CRM data and actions — promising large productivity and automation gains but also raising governance, data‑access, cost, and reliability questions; the trend shifts buying and integration models (marketplace agent acquisitions, per‑agent pricing, multi‑cloud agent orchestration) across Oracle, Salesforce, AWS, OpenAI and specialist startups. (aws.amazon.com)

Key players include incumbents and cloud providers (IBM partnering with Oracle and building watsonx Orchestrate agents; Oracle operating the Oracle Fusion AI Agent Marketplace; AWS promoting Amazon Bedrock Agents and patterns to integrate with Salesforce Agentforce), platform makers and integrators (Salesforce/Agentforce and OpenAI Agent Builder + MCP ecosystem), CRM vendors (HubSpot), AI model/platform vendors (OpenAI, Anthropic), middleware/MCP specialists (Composio, Rube), startups (Aurasell with a $30M seed) and investors (Next47 led Aurasell’s round). (newsroom.ibm.com)