How large organisations are buying, building, and deploying AI: Fortune 500 case studies, vendor partnerships, platform launches, and the workflow changes shipping inside companies right now.
Topic brief
Brief updated Aug 6, 2026
Enterprise AI is the adoption of AI models, agents and automation inside businesses, spanning productivity copilots, customer-experience tools, back-office automation and custom agentic applications. It is where frontier-model capability meets the constraints of real organizations: legacy systems, governed data, identity and permissions, security review, procurement and measurable return on investment. For practitioners, the defining challenge is not model quality alone but turning capability into governed, reliable production workflows.
The stakeholders span the whole organization. Engineering and platform teams integrate models and manage governance, security and cost. Line-of-business leaders chase productivity and customer-experience gains. Executives weigh build versus buy and vendor risk. A dense vendor ecosystem of frontier labs, cloud providers, systems integrators, agent-platform startups and governance-tooling makers competes to supply the stack.
The topic matters because enterprise AI is where the technology's economic impact is realized or stalls, and the evidence about that arrives in two incompatible forms. Aggregate surveys repeatedly report weak or unmeasurable return and low organizational readiness. Firm-level accounts report large gains. Both can be true at once, and reconciling them is the analytical task this hub exists to support. Underneath sit the recurring structural themes: multi-vendor sprawl, identity-aware access, security drift in connectors that change after approval, durable agent state and memory, token cost and pricing models, and the organizational change management that decides whether a pilot becomes a durable system.
The newest evidence splits cleanly into two stories that only look contradictory. On the revenue side, the numbers are large and concentrated: Palantir reported second-quarter 2026 revenue of 1.935 billion dollars, up 93% year over year, with US commercial revenue at 764 million dollars, up 149%, and US government revenue at 809 million dollars, up 90%, and raised full-year guidance to between 8.150 and 8.158 billion dollars. OpenAI said on July 31 that its models reach more than 1 billion active users and more than 2 million businesses, without specifying a measurement window and without independent audit. Yellow.ai agreed on August 3 to a Bluerock Acquisition Corp. business combination the companies put at roughly 550 million dollars in pro forma equity value, though the same-day SEC filing values the merger consideration at 300 million dollars and closing remains subject to shareholder approval. On the adoption side, the picture is thinner: Booking Holdings reported second-quarter revenue of 7.4 billion dollars, up 8%, and gross bookings of 51 billion dollars, up 9%, while PYMNTS reported that traffic from large language models remained below 1% of room nights and that Booking was applying AI inside customer service and supplier workflows rather than at the top of the funnel. A Box-commissioned Harris Poll of 1,640 IT decision-makers, conducted April 30 to May 8, found 83% saying their organizations were running AI agents while only 36% of those using or testing agents had connected them to trusted internal content across many use cases, 49% reported an AI-related data-exposure incident, and only 34% said they had formal standards governing agent access to company data. Those figures are vendor-commissioned, unweighted and self-reported, but they describe the same gap the product releases are chasing.
Which is the second story: the control plane is now something you buy. Cloudflare open-sourced Cloudflare OS on August 5 under Apache 2.0, a browser-based agent workspace it had run internally that pairs organization-curated context and an isolated code runtime with a governance layer using Gatekeepers and Cloudflare Access, released as early access. The same day it detailed dollar-based spend limits in AI Gateway, in open beta, with identity-driven budgets that attach verified user, group or service-token context via Cloudflare Access still in closed beta and able to block requests or route them to a cheaper fallback model. 1Password launched Privileged Access on July 28, provisioning task-scoped permissions for people, service accounts and AI agents in target systems and removing them at session end. Cyera announced a 1 billion dollar agreement to acquire non-human identity company Oasis Security, which Globes describes as currently a letter of intent subject to a binding agreement. Ethyca launched Astralis on August 4 for real-time purpose-based access checks on agent data paths, running inside the customer's cloud. CrowdStrike connected Falcon AI Detection and Response to Microsoft Copilot Studio and Claude Code on July 30, with the important caveat in its own documentation that the Copilot integration evaluates only tool name and input parameters and that the Claude Code collector cannot see model responses, system prompts or subagent context. Microsoft moved on two fronts, refining Restricted Content Discovery so recently accessed files from protected SharePoint sites stop resurfacing through Search and Copilot discovery, and putting a Purview DLP control that excludes externally received email from Copilot grounding into preview with general availability listed for January 2027. The motivating threat is concrete: Barracuda published a controlled proof of concept on August 4 in which an attacker already holding an employee mailbox used Copilot to find targets, imitate colleagues and escalate to a CEO account, redirecting a simulated 247,500 dollar wire transfer. Underneath all of it, cost discipline stayed unresolved. Reports published July 30 said a failed Amazon Claude Sonnet deployment cost about 1.8 million dollars, ran 860% over budget and went undetected for five months, with Amazon calling the cases isolated learning examples and saying it was developing automated spending guardrails, while Harness's vendor-sponsored 2026 State of AI in FinOps survey of 700 respondents estimated 26% of AI spending wasted, 52% with no clear cost owner and only 20% able to diagnose a doubled bill within hours. OpenAI's price cuts, to 0.20 dollars per million input tokens for GPT-5.6 Luna and 2 dollars for Terra, and ABC's reporting on Australian companies routing simpler work to open-weight Chinese models after Moonshot AI published Kimi K3's weights on July 27, both push in the same direction without solving attribution.
The most checkable signals here are dated availability and deal conditions. Cloudflare OS is labeled early access and identity-driven AI Gateway budgets remain in closed beta, so watch whether either reaches general availability and what connectors, authorization model, observability and data-residency options ship with it. Microsoft lists general-availability rollout of the Copilot external-email DLP control starting in January 2027, and ET TravelWorld said Amex GBT's Egencia connector for Claude is planned for Egencia customers in Q3 2026. On transactions, watch whether Nscale closes the Anyscale acquisition in the second half of 2026 as announced and whether any price is ever confirmed, since the companies did not disclose terms and the 1.65 billion dollar figure comes from secondary coverage; whether Cyera's Oasis Security letter of intent converts to a binding agreement; whether Bloomberg publishes a closing date or terms for Canoe Intelligence; and whether Yellow.ai's SPAC merger clears shareholder approval given the gap between the 550 million dollar pro forma equity value and the 300 million dollar merger consideration in its SEC filing. On cost, watch whether Amazon publishes anything about the automated spending guardrails it said it was developing, because the available reporting does not break the overruns into model tokens, retries, engineering errors or other infrastructure. On regulation and provenance, BaFin says monitoring of high-risk AI systems including certain creditworthiness uses begins in December 2027, and the House committees' July 31 document request to DoorDash about its use of Moonshot AI's Kimi K2.6 alleges no wrongdoing but is worth following for what model-provenance documentation regulators expect. Also watch whether CrowdStrike closes the visibility gaps its own documentation names for Copilot Studio and Claude Code, whether Torq publishes methodology, sample sizes and false-positive rates behind its vendor-reported 85% and later above-91% agreement figures, and whether Uber's newly formed dedicated team reports error rates, review requirements and costs alongside its company-reported Agentic Pods workflow times.
Because deploying an agent and connecting it safely to company data are separate projects, and only the first has a champion. A Box-commissioned Harris Poll of 1,640 IT decision-makers in four countries found 83% saying their organizations ran AI agents, but only 36% of those using or testing agents had connected them to trusted internal content across many use cases, 49% reported an AI-related data-exposure incident, and just 34% had formal standards governing agent access to company data. The survey is vendor-commissioned, unweighted and self-reported, so treat the exact percentages as directional. The pattern it describes is what the identity, budget and DLP products launched in this window are selling against.
Missing attribution rather than headline token prices. Reports published July 30 said a failed internal Amazon deployment of Claude Sonnet cost about 1.8 million dollars, ran 860% over budget and went undetected for five months, with roughly 541,000 dollars on a financial-auditing tool and 134,000 dollars on a delivery-speed task; Amazon called these isolated learning examples and said it was developing automated spending guardrails. Harness's vendor-sponsored 2026 State of AI in FinOps survey of 700 respondents across five countries estimated 26% of AI spending wasted, with 52% lacking a clear AI-cost owner, 72% having hit a surprise bill in the past year and only 20% able to diagnose a doubled bill within hours. Falling list prices do not fix this, which is why Cloudflare's dollar-based AI Gateway limits and identity-attached budgets matter more than a per-token comparison.
No, and the vendors say so. CrowdStrike's Falcon AI Detection and Response connects to Microsoft Copilot Studio and Claude Code, and the Copilot integration can allow or block a proposed tool invocation before execution, but CrowdStrike's documentation states it evaluates only the tool name and input parameters, with Copilot output and conversation context outside its scope. The Claude Code collector uses native hooks to evaluate submitted prompts and pre- and post-tool events without installing a separate binary, but cannot see model responses, system prompts or subagent context. Plan controls around those blind spots rather than assuming end-to-end coverage.
Barracuda published a controlled proof of concept on August 4 that is worth reading precisely because of its preconditions. The scenario begins with an attacker who already controls an employee mailbox, then uses Microsoft Copilot to identify targets, imitate trusted colleagues and escalate to a CEO account, redirecting a simulated 247,500 dollar wire transfer. It was a laboratory demonstration, not a disclosed customer breach or real loss. The practical implication is that the assistant accelerates use of data and privileges the compromised account already had, which makes mailbox-rule, session and payment-workflow monitoring the relevant controls rather than assistant-specific filtering alone.
Persistence and governance claims that have not been benchmarked. MinIO's AIStor Memory, introduced July 29, separates durable organizational knowledge, active workspace state and credential custody into distinct services with different access boundaries, and says Agent Biography can preserve evidence, decisions, corrections, provenance, outcomes and unfinished work across runs; MinIO also cautions that it does not change the underlying model or guarantee perfect recall, and the announcement includes no pricing, retrieval benchmarks or latency data. Similarly, Guidewire's Qusar Agentic Framework became generally available on August 3, but availability varies by feature and its development-speed figures come from internal benchmarks. Ask for exception rates, permission boundaries and audit trails in your own workflows.
The evidence points the other way. Okta's Enterprise AI Index, covering sanctioned application access at more than 20,000 organizations through June 2026, found that most organizations used multiple AI platforms and that single-provider use fell 1.2 percentage points in June. Cost pressure is pulling in the same direction: ABC reported Australian companies routing simpler workloads to cheaper models after Moonshot AI published Kimi K3's weights on July 27, and OpenAI cut list prices for GPT-5.6 Luna and Terra. Multi-vendor deployment is also what makes the identity and gateway layer commercially interesting, since budgets, permissions and audit trails have to span providers rather than sit inside one.
Latest coverage
Showing newest first with full-topic pagination
Payhawk made Agent Fetch generally available to all customers on August 22, 2026, adding automated vendor-portal invoice retrieval to its finance platform. The company reports that the agent retrieves invoices 12 times faster than humans and targets a reduction in average submission time from 23.8 days to two days. More than 550 businesses activated the feature during beta, The Fintech Times reported.
Coupa's Payment Batch Creation Agent prepared 14 payment batches covering 2,395 payments worth $20.1 million during its first five weeks in production, PYMNTS reported on August 21. The agent prepares approved invoices for payment but does not authorize the release of funds; a human or separate system retains that approval step.
Clifford Chance launched an internal AI-enabled knowledge management platform in July, developed with Epiq Advisory for Law Firms and Microsoft. Global Legal Post reports that the library contains more than 400,000 curated, access-controlled documents and includes permissioned search designed to preserve client confidentiality. The system runs within the firm's Microsoft 365 and Azure environment.
Credit unions reached 46% chatbot adoption in 2026, up from 3% in 2019, according to Cornerstone Advisors data cited by PYMNTS and CU Daily. PYMNTS Intelligence research conducted with Velera found that 75% of small and midsized businesses would use at least one AI feature from their financial institution within two years, with demand centered on financial guidance rather than autonomous decision-making.
NTT DATA and Palo Alto Networks announced a multiyear alliance on August 20 to help organizations adopt AI securely and modernize cybersecurity. The companies said it targets $1 billion in joint business by 2029 and will begin with six transformation areas, including AI governance, agentic security operations, identity, cloud security, secure access and firewall modernization.
Expedia Group eliminated eight senior product and technology leaders in an AI-focused reorganization reported by Skift on August 20. GeekWire reported that Expedia confirmed the departures and that the new structure uses smaller end-to-end squads, while Skift reported that AI, data, and agentic platforms are centralized under Chief AI and Data Officer Xavier Amatriain.
Twin1 AI emerged from stealth on August 20 with a $20 million seed round to develop AI-powered digital twins for knowledge workers. City AM reports that Bessemer Venture Partners, Tribeca Venture Partners, and Aramco Ventures co-led the financing, while Linklaters, Orrick, and Dechert are among the company's named law-firm customers. The platform connects employee context from workplace systems to answer questions, draft communications, and perform selected tasks.
Slack launched Slack Code on August 20, 2026, creating project-specific channels where teams can assign coding agents work and review it together. The Verge reports that participants can inspect code diffs, view live HTML previews, provide feedback, and approve work before shipment. The Register reports that completed channels can archive automatically while retaining a searchable record.
Ant International launched Falcon Time-Series Transformer Model 2.0 on August 20, with Reuters reporting that Citi, HSBC, Deutsche Bank, Standard Chartered and Barclays are among six large banking partners. The specialized model is designed for foreign-exchange forecasting and liquidity-risk management in cross-border payments. Ant executive Kelvin Li told Reuters that precise forecasting can reduce FX hedging and allocation costs by more than 60%.
Deloitte's 2026 research finds that only about one in five organizations are prepared to redesign business processes for autonomous AI agents, HR Dive reported on Aug. 20. Among 501 surveyed U.S. senior managers and C-suite executives, 61% expected most agents to be largely autonomous with humans primarily providing oversight. Deloitte separately found just 21% of global respondents had mature agentic AI governance.
Cloudera launched Anywhere Cloud on August 19, providing a control plane for data and AI workloads across public clouds, sovereign infrastructure, and on-premises environments. StorageReview and Techzine report that the modular platform supports services including Spark, Kafka, and Trino without requiring data migration, while applying centrally managed governance and zero-trust controls. Cloudera frames the release around production deployment of agentic AI on distributed enterprise data.
Forrester introduced its AI Disruption Model on August 19, assessing AI's likely effect across 17 technology and service categories containing more than 200 markets. The firm identifies infrastructure, data and AI, and identity and security as broad growth areas, while reporting greater exposure for labor-intensive services including software development and technology implementation.
Meta released a beta Meta AI app for Mac on August 19, adding screen sharing and system-wide dictation. Axios and MacRumors report that the app targets creators and small businesses with Facebook and Instagram analytics plus Google Workspace connections; Workspace access requires a professional Facebook or Instagram account.
Hyundai Motor Group says it cut the time needed to search and review crash-test data by 90% after applying AI to historical engineering records. The company reported the result on August 12 as part of a wider AI program; it points to faster engineering analysis, while the underlying safety decisions still require separate technical validation.
Paul Stokes, co-founder and chief executive of Prevalent AI, answered six technical questions from Lets Data Science after the London company announced a $22 million growth investment. He set out how the platform arbitrates between data sources that disagree, using fidelity rules, frequency weighting and per-field precedence, and said the system is designed to return a transparent unknown rather than a clean but incorrect answer. On sovereignty he was candid: the platform supplies the trusted data layer, and customers reach full data sovereignty only by running open-weight models on their own infrastructure. Revenue and customer numbers remain undisclosed.
Prevalent AI raised $22 million from Integrity Growth Partners, its first outside capital since the London company was founded in 2017. The August 19 growth investment will fund U.S. expansion, a formal go-to-market organization, and work to extend its enterprise data fabric beyond cybersecurity into compliance, financial crime, and operational risk.
QualityKiosk Technologies opened an engineering hub in Hyderabad's HITEC City on August 19, focused on AI reliability engineering, AI assurance, agentic engineering, product engineering, CloudOps, and automation. Urban Acres reports that the facility supports an existing local workforce of more than 40 employees, while Devdiscourse reports that the office also houses marketing, branding, and strategic-relations functions.
Mistral says its indexed Google Drive and SharePoint Knowledge Connectors will be removed at the end of August 2026 in favor of MCP Connectors. Administrators must add the replacements and users must reconnect their accounts, while Mistral's public documentation leaves the replacement retrieval architecture and the old index's cutoff-time deletion schedule unresolved.
Personetics and Plaid announced a partnership on August 18 that integrates open finance data into Personetics' AI-powered personalization platform for banks and credit unions. According to PYMNTS' reporting on the companies' release, the integration combines an institution's internal account data with externally held account information to generate personalized insights and recommended next actions in digital banking channels.
Microsoft's Business Central 2026 release wave 1 put an Admin Center MCP server into public preview on April 1, 2026. The company's documentation says MCP-compatible AI agents can retrieve environment information and propose actions such as sandbox copies or update scheduling through standardized administration APIs, with explicit user confirmation retained for follow-up actions.
Rippling President and CPO Matt MacInnis said on July 30 that the company ran roughly 2,100 scored attempts per model across 15 AI models on payroll, personnel, and finance tasks. SaaStr's August 18 analysis found the leading models clustered within a few accuracy points while differing sharply in cost and latency, underscoring that model selection depends on the workload rather than one overall winner.
NLPatent rebranded as Clerq on August 18 and launched an agentic AI platform for patent research workflows. The company introduced rapid invention-disclosure triage and full patentability reports with cited references, while BetaKit reported that Clerq is targeting legal teams facing more complex applications and patent backlogs. Clerq's claim that the work can take roughly 10 minutes has not been independently tested.
Synchrony announced an enterprise collaboration with OpenAI on August 17 and said it will launch a ChatGPT plugin for discovering savings, promotional financing, deals, and offers from participating partners. The company also said it plans to deploy GPT-5.6 Sol, Terra, and Luna through ChatGPT Work, Codex, and AWS Bedrock, while its release gives conflicting signals on whether the plugin is already available.
HCLTech and NetApp expanded their partnership on August 13 with a hybrid cloud storage-as-a-service offering for enterprise AI and other data-driven workloads. The companies say it combines HCLTech's U4X infrastructure framework with NetApp Keystone's pay-as-you-go storage and connects with HCLTech's AI Factory suite; pricing and underlying data services were not disclosed.
Aderant announced general availability of an iTimekeep integration with Harvey on August 17, enabling work performed in Harvey to create draft time entries in iTimekeep. LegalTechnology.com reports that entries include matter context, duration, and an AI-generated narrative for lawyers to review, edit, and submit through their firms' existing approval processes.
A July 2026 Carnegie Mellon University-Larridin working draft found that companies with the most concrete AI disclosures were associated with 8.0 percentage points higher year-over-year revenue growth than those with the least concrete disclosures after sector, size, and prior-growth controls. The cross-sectional study did not find comparable margin gains or risk-adjusted stock-return predictability, and its authors caution that the result is an association rather than evidence that AI caused the growth.
EY is building an AI Value Realization Office to centralize AI spending and measure business impact. EY Global Chief Innovation Officer Joe Depa said the firm created the office to steer model selection and usage governance, while Business Insider reported on August 15 that the function should be fully operational within a couple of months.
OpenAI has expanded ChatGPT's Google Drive integration so eligible paid users can open and edit native Google Docs, Sheets and Slides without leaving ChatGPT, with changes written back to the underlying Drive file. Availability depends on the plan, surface, file type and workspace configuration, and administrators may need to approve the Google OAuth scopes required for write actions.
Decagon has crossed $100 million in annualized revenue and argues that enterprise AI software should not require long-term embedded engineers, CEO Jesse Zhang told Newcomer in an August 11 interview. The company still uses forward-deployed engineers to launch customers, but says its product should leave client teams able to configure and expand AI support workflows themselves.
WRITER released Palmyra X6, a new flagship model, alongside a rebuilt Agent harness and governance tools on August 13. The company reports that the combined model and harness reduce average agent costs by 52%, improve speed by 48%, and raise quality by 10% versus prior performance. The release targets enterprise teams managing token spend across multi-step AI workflows.
