Startup Spotlight: 5 AI Companies to Watch in the Enterprise
Five companies. Five different ways AI is being deployed across the enterprise. Together, they map the emerging shape of what happens when artificial intelligence moves from feature to foundation.
Global AI spending is on track to reach $2.59 trillion in 2026, with AI software alone accounting for more than $453 billion and AI infrastructure representing over 45 percent of total spend. At those scales, AI is no longer just a feature inside applications — it's becoming the operating layer for how companies secure cloud environments, run models, automate professional work, deploy physical labor, and manage financial operations. This month's special identifies five companies that each represent a different layer of that emerging stack, and explains why each matters for what comes next.
Why AI-Powered Enterprise Matters Now
MARKET CONTEXTEnterprise AI has moved from pilot projects to production infrastructure. Gartner forecasts worldwide AI spending of $2.59 trillion in 2026, up 47 percent from 2025, with AI software projected at $453.6 billion and AI cybersecurity alone expected to reach $51.3 billion — nearly double the prior year. AI infrastructure is anticipated to account for more than 45 percent of that total, reflecting the shift from experimentation to scaled deployment across servers, networking, and cloud environments.
This matters because AI is becoming the layer on which other enterprise capabilities are built. Security tools now protect AI workloads alongside traditional infrastructure. Inference platforms let companies run open models on flexible, global infrastructure rather than relying solely on hyperscaler APIs. Vertical AI companies embed deeply into domain-specific workflows, accumulating proprietary data and practices that horizontal tools can't match. Physical AI systems are moving from lab demos to early production deployments in real factories and warehouses. Financial operations platforms are deploying AI agents to automate high-volume, repetitive back-office work under clear policy guardrails.
The five companies below — Wiz, Together AI, Harvey, Figure AI, and the Ramp/Brex pair — illustrate distinct but interconnected layers of this emerging stack. None of them is trying to be everything to everyone. Each is betting that depth in a specific layer will matter more than breadth across all of them.
The Benchmark — Wiz
AI-POWERED CYBERSECURITYWiz has become the reference point for what AI-assisted enterprise security looks like in practice. In March 2026, Google completed its acquisition of Wiz, and the platform joined Google Cloud while retaining its brand and multi-cloud mandate across AWS, Azure, Google Cloud, and Oracle Cloud. Rather than remaining a pure cloud security posture management (CSPM) vendor, Wiz has evolved into what it calls an AI Application Protection Platform (AI-APP), extending its graph-based risk engine from infrastructure into AI workloads.
Wiz's core innovation is the Security Graph, which connects code, cloud configurations, identities, data, and runtime signals into a single contextual map of attack paths. This graph underpins traditional CNAPP capabilities — CSPM, cloud workload protection, CIEM, DSPM, and data detection and response — but now also covers AI assets such as models, training data, AI agents, and Model Context Protocol (MCP) servers. At RSAC 2026, Wiz introduced AI-APP as the next step beyond CNAPP, adding visibility, risk analysis, and runtime protection specifically for AI applications.
In practical terms, Wiz's AI-SPM module scans for exposed AI models, unsecured training data, and risky AI service configurations across cloud accounts. [8] Its agentic security model deploys specialized Red, Blue, and Green AI agents that autonomously investigate threats, validate exploitability, and remediate risks, paired with Wiz Workflows that automate response directly in the tools security teams already use. Integrations with partners like JFrog shrink the time between risk detection and verified code fixes by connecting Wiz's runtime visibility to the software supply chain.
What Wiz reveals about enterprise AI: Security is becoming inseparable from AI deployment. As companies embed AI agents and models into production systems, the attack surface expands beyond traditional infrastructure to include AI-specific risks like prompt injection, model abuse, and data leakage. Wiz's trajectory — from cloud security leader to AI-APP pioneer — signals that the next phase of enterprise AI will be defined not just by capability, but by how securely and governably those capabilities operate at scale.
The Infrastructure Play — Together AI
AI INFRASTRUCTUREIf Wiz secures the AI layer, Together AI helps companies build and run it. Together AI's role is to provide the infrastructure and tooling that enterprises need to build, fine-tune, and deploy open and customized AI models at scale. In September 2026, Together AI announced a partnership with Equinix and NVIDIA to launch the Equinix Inference Exchange, an AI inference platform running across Equinix's global data centers.
The Equinix Inference Exchange combines NVIDIA Enterprise Reference Architectures with Together AI's inference platform and Equinix's colocation footprint to optimize deployment speed, flexibility, and compliance for enterprise AI workloads. Together AI's platform supports more than 200 open-source models and offers both multi-tenant deployments for shared efficiency and dedicated single-tenant environments for workloads that require isolated capacity. This architecture matters because it lets enterprises place inference closer to their data, applications, and users, cutting latency and simplifying compliance in regulated industries.
Together AI's value proposition sits between hyperscaler AI services and do-it-yourself open-source stacks. It gives companies access to open-model flexibility and choice without requiring them to manage the full complexity of model hosting, scaling, and optimization. For organizations that want to avoid vendor lock-in or need to run specialized models for specific domains, Together AI provides a middle path: open models, enterprise-grade infrastructure, and deployment options that can be tuned to performance, cost, and compliance requirements.
What Together AI reveals about enterprise AI: The infrastructure layer is becoming a strategic battleground. As AI adoption expands, companies will increasingly demand choice over which models run, where they run, and how they are governed. Together AI's positioning — open models on flexible, global infrastructure — suggests that the next wave of enterprise AI will be less about a single dominant model provider and more about a diversified, multi-model, multi-cloud ecosystem where infrastructure partners enable choice and control.
The Vertical Disruptor — Harvey
LEGAL AIHarvey is the clearest example so far of a vertical AI company turning domain expertise into a defensible platform. Harvey develops legal AI tools for law firms and corporate legal departments, automating workflows such as contract analysis, due diligence, compliance, and litigation support. In September 2026, Harvey raised $550 million at a $15.5 billion valuation — up roughly 41 percent from its $11 billion mark just six months earlier — signaling strong investor conviction in vertical AI's long-term value.
Harvey's differentiation lies in its domain-specific approach. Rather than offering generic access to large language models, Harvey builds software that remembers institutional practices, coordinates legal workflows, and keeps lawyers in control. In August 2026, Harvey unveiled Tenet, its first proprietary post-trained legal AI model, benchmarked at frontier level for a fraction of typical API costs. Tenet represents a strategic shift: Harvey is moving from orchestrating third-party models to owning its own legal-specific AI stack, which could improve performance, reduce costs, and deepen moats around its workflows.
Revenue disclosures tied to Harvey's latest round indicate approximately $300 million in annual revenue, implying a valuation multiple of roughly 51.7x. While high, this multiple reflects the market's bet that vertical AI platforms in high-value professions can command premium pricing and sticky adoption if they demonstrably improve productivity and quality in core workflows.
What Harvey reveals about enterprise AI: Vertical AI companies may be where the most durable enterprise value is created. Horizontal AI tools are powerful, but they often compete on price and features in crowded markets. Vertical players like Harvey embed deeply into domain-specific workflows, accumulate proprietary data and practices, and build models tailored to those contexts. Harvey's rapid valuation growth suggests that investors see vertical AI — not just horizontal platforms — as a central driver of the next phase of enterprise AI adoption.
The Physical AI Bet — Figure AI
HUMANOID ROBOTICSFigure AI represents the most ambitious bet in this set: combining advanced AI models with humanoid robotics to create a general-purpose physical labor platform. Figure has compressed robot development, AI training, factory construction, and fundraising into four years, with its BMW deployment serving as the most meaningful public evidence of progress.
Figure 02, the company's previous-generation humanoid, completed an extended production deployment at BMW Group's Spartanburg plant in 2025. During that deployment, Figure 02 logged more than 1,250 cumulative hours, handled over 90,000 stamped parts, and contributed to the production of more than 30,000 BMW X3 vehicles while operating 10-hour shifts from Monday through Friday. In November 2025, Figure announced it was retiring the Figure 02 fleet following the introduction of Figure 03, meaning the BMW deployment represents a completed phase rather than an ongoing large-scale rollout of that generation.
Figure 03, introduced in October 2025, is designed as a general-purpose platform around Figure's proprietary Helix vision-language-action system. As of June 2026, approximately 40 Figure 03 robots have been deployed in Hall 52 of the BMW Spartanburg plant for a logistics sequencing workflow, billed at approximately $25 per hour. Tasks have evolved from simple pick-and-place of stamped parts to more variable logistics sorting, picking from unsorted parts, and delivering them to line workers in assembly order. The figure says its BotQ factory has delivered more than 350 Figure 03 units and demonstrated a one-robot-per-hour assembly cycle, though the company notes that units are divided among internal research, data collection, housework development, and commercial-use-case development.
Beyond BMW, Figure has announced a commercial agreement with Catalyst Brands beginning at a Reno distribution center, and Figure 02 has commenced large-scale pilot deployments across three major logistics facilities in North America and Europe. The company has also confirmed plans to establish pilot operations in India by Q3 2025, with preliminary discussions underway with major Indian automotive and electronics manufacturers for repetitive assembly, quality inspection, and material handling tasks.
What Figure AI reveals about enterprise AI: Physical AI is moving from lab demos to early production, but commercial scale remains unproven. Humanoid robots are now performing real work in real factories, but deployments are still limited to specific stations and workflows rather than broad, multi-site rollouts. Figure's trajectory suggests that the next phase of enterprise AI will include not just digital agents and models, but embodied agents that can operate in physical environments — though the path to widespread adoption will depend on reliability, cost, and integration complexity.
The Fintech Wildcard — Ramp / Brex
AI FINANCIAL OPERATIONSRamp and Brex both started as corporate card and expense management platforms, but each is now positioning itself as an AI-native finance operating system. The question is not which company will "win," but how each is using AI to eliminate repetitive operational work and what that implies for the future of enterprise finance.
Ramp: AI agents across procure-to-pay. Ramp brings together corporate cards, expense tracking, bill pay, and procurement in a single finance operations platform, then applies AI agents to the repetitive work running through each of those areas. In September 2026, Ramp announced a strategic collaboration agreement with AWS, making its AI-enabled finance platform available on AWS Marketplace. Through this collaboration, Ramp's AI agents operate within the platform to code expenses, enforce spending policy, and process invoices on an ongoing basis.
Ramp Procurement runs the entire procure-to-pay process through AI agents: employees describe what they need, the Procurement Agent handles vendor sourcing, due diligence, and compliance checks, and approved requests flow straight into purchase orders and Bill Pay with no manual handoffs. The platform includes custom intake forms that route different purchase categories through different approval workflows, and AI agents that handle work once reserved for dedicated headcount — from sourcing vendors to compliance checks to renewal prep. Ramp reports powering over $200 billion in annual purchase volume across more than 70,000 organizations, with its AI agents intended to provide continuous automation for back-office tasks while ensuring every action remains auditable and reversible under enforceable financial guardrails.
Brex: AI-native spend management and expense automation. Brex originated as a venture-backed corporate credit card for high-growth tech startups and has expanded into Brex Empower, a global financial operating system offering high-limit corporate cards, multi-currency local accounts in 50+ countries, automated expense management, travel booking, and venture treasury vaults. Brex's AI strategy centers on automating expense and accounting workflows end-to-end. When someone swipes a corporate card, the platform automatically collects and matches receipts while AI helps fill in purchase details and memos.
Brex uses AI agents and automation to reduce manual work across three groups: For employees, AI collects and matches itemized receipts automatically, creates memos, and populates expense details with less manual entry. For managers, intelligent approval routing sends expenses directly to the right decision-makers based on corporate hierarchy, project budgets, or merchant types. For finance teams, AI assigns transactions to the right accounting category, project, or department while spotting policy questions or potential issues earlier; routine, compliant expenses move through review automatically, with human attention reserved for exceptions.
Brex AI and direct data feeds with merchants like Lyft, Uber, DoorDash, and Amazon Business enable automatic capture of line-item receipt metadata, and the company reports that more than 80 percent of expense processing is handled via AI automation. Brex also exposes its expense data via API and Model Context Protocol connectors, allowing companies to integrate AI agents directly with Brex expense data for custom workflows.
Different approaches, shared direction. Ramp and Brex are converging on a similar end state — AI-driven finance operations — but from different angles. Ramp emphasizes end-to-end procure-to-pay automation with AI agents that act across intake, sourcing, compliance, PO generation, and payment, positioning itself as a full finance operations platform.
Brex emphasizes AI-native expense and spend management, embedding automation deeply into card swipes, receipt matching, approval routing, and accounting categorization, while also opening its data to external AI agents via API and MCP.
What Ramp and Brex reveal about enterprise AI: Financial operations are becoming a primary target for AI agents. Both companies show that the most immediate value of AI in the enterprise may not be in flashy new products, but in automating high-volume, repetitive back-office work that currently consumes significant human time. Their trajectories suggest that the next phase of enterprise AI will include not just chat-based assistants, but autonomous agents that can execute multi-step workflows under clear policy guardrails — starting with finance, then expanding to other operational domains.
What These Companies Have in Common
PATTERNSDespite operating in different layers, these five companies share several characteristics:
- AI as infrastructure, not just a feature: Each company treats AI as a core operating layer — whether securing AI workloads (Wiz), running open models (Together AI), automating legal work (Harvey), deploying physical labor (Figure AI), or managing financial operations (Ramp/Brex).
- Domain-specific depth: Success depends on deep integration into specific workflows. Wiz's Security Graph, Harvey's Tenet model, Figure's Helix system, and Ramp/Brex's finance platforms all reflect a move from generic AI to specialized, context-aware systems.
- Agents and automation: All five are moving toward agentic models — AI systems that can investigate, decide, and act within defined guardrails. Wiz's Red/Blue/Green agents, Ramp's Procurement Agent, Brex's expense automation, and Figure's embodied agents all point to a future where AI does work, not just analysis.
- Enterprise-grade requirements: Each company must meet high bars for security, compliance, auditability, and integration. Wiz's multi-cloud coverage, Together AI's dedicated environments, Harvey's workflow controls, Figure's production deployments, and Ramp/Brex's policy guardrails all reflect the realities of enterprise adoption.
What Could Go Wrong
RISKSSeveral risks could slow or reshape this AI-powered enterprise trajectory:
- Security and governance gaps: As AI agents gain more autonomy, the risk of misconfiguration, data leakage, and unintended actions grows. Wiz's emergence as an AI-APP vendor underscores that many organizations are not yet equipped to secure AI workloads at scale.
- Overhyped valuations vs. proven scale: Harvey's $15.5 billion valuation and Figure's ambitious production targets run ahead of proven commercial scale. If vertical AI or humanoid robotics fail to deliver expected productivity gains, valuations could compress sharply.
- Infrastructure bottlenecks: AI infrastructure already accounts for more than 45 percent of AI spending, but supply constraints, cost pressures, and integration complexity could limit how quickly enterprises can scale AI deployments.
- Regulatory and liability questions: As AI agents take on more responsibility in legal, financial, and physical workflows, questions around accountability, liability, and compliance will intensify — particularly for vertical AI and physical AI where errors can have direct real-world consequences.
What to Watch Next
STRATEGIC PRIORITIES- Wiz's post-acquisition roadmap: How Google integrates Wiz into its broader cloud and AI security portfolio, and whether AI-APP becomes a standard layer for securing enterprise AI across hyperscalers.
- Together AI's enterprise deployments: The pace and scale of Equinix Inference Exchange adoption, and whether open-model infrastructure becomes a mainstream alternative to hyperscaler AI services.
- Harvey's model and workflow expansion: Whether Tenet and future proprietary models translate into measurable productivity gains and deeper workflow integration, and how competitors respond with their own vertical AI stacks.
- Figure's commercial milestones: Progress beyond pilot deployments — particularly additional named customers, multi-site rollouts, and clearer unit economics for humanoid robotics in manufacturing and logistics.
- Ramp and Brex's agent capabilities: The scope of autonomous workflows each platform can handle, adoption of AI-driven procure-to-pay and expense automation, and whether finance teams begin to treat AI agents as core operational infrastructure.
AI-Powered Enterprise at a Glance
| Layer | Company | Focus |
| Security | Wiz | AI Application Protection |
| Infrastructure | Together AI | Open-Model Inference |
| Vertical AI | Harvey | Legal Workflows |
| Physical AI | Figure AI | Humanoid Robotics |
| Financial Operations | Ramp / Brex | AI Finance Agents |
The next phase of enterprise AI will be defined less by which model is "best" and more by how AI is secured, deployed, specialized, embodied, and operationalized. Wiz shows that AI security must evolve alongside AI adoption. Together, AI highlights the strategic importance of open, flexible infrastructure. Harvey demonstrates the power of vertical AI in high-value professions. Figure AI tests whether embodied AI can become a scalable labor platform. Ramp and Brex illustrate how AI agents can transform back-office operations.
Together, these five companies map a coherent picture of AI-powered enterprise: a stack where security, infrastructure, domain expertise, physical automation, and operational workflows are all being rewritten around AI.
The companies that thrive in this environment will be those that can combine deep domain knowledge with robust, agentic automation — and do so in ways that are secure, governable, and economically compelling.
AI-Powered Enterprise: 5 Companies
Security → Infrastructure → Vertical AI → Physical AI → Financial Operations
Information Sources:
- Reuters — "Legal AI startup Harvey reaches $15.5 billion valuation in new funding round," September 9, 2026
- Dito / Google Cloud — "Unifying Code, Cloud, and AI Security with Wiz," September 16, 2026
- Nightfall AI — "Wiz Reviews 2026," September 9, 2026
- Equinix Newsroom — "Equinix Accelerates AI Inference for Enterprises with NVIDIA and Together AI," September 2, 2026
- PRNewswire — "EPAM Partners with Wiz to Help Enterprises Reduce Cloud Risk," August 25, 2026
- Channel Life — "Equinix launches AI inference exchange with Nvidia," September 3, 2026
- Multiples.vc — "Harvey Revenue, Valuation, Funding & Investors," September 19, 2026
- Tech Insider — "Wiz vs Orca Security vs Prisma Cloud CSPM 2026," September 5, 2026
- Entrepreneur Loop — "Harvey AI Valuation Jumps to $15.5B — $550M Round Explained," September 10, 2026
- PRNewswire — "EPAM Partners with Wiz to Help Enterprises Reduce Cloud Risk and Strengthen Cyber Resilience," August 25, 2026
- Together AI (X) — "Open source is becoming the default way enterprises build with AI," September 2, 2026
- PeerSpot — "Wiz reviews 2026," February 24, 2026
- Remio.ai — "Harvey AI Funding Hits $550 Million as Legal AI Becomes a Platform War," September 11, 2026
- Business Wire — "JFrog Partners with Wiz to Close the Gap on AI-Era Threats," September 2, 2026
- Enterprise DNA — "Harvey Builds Its Own AI Model for Law Firms," August 22, 2026
- Monday.com — "15 best AI agents for finance teams in 2026," August 28, 2026
- Ramp — "AI in Procurement: Use Cases, Benefits, & Implementation," August 28, 2026
- Ramp — "Best Procurement Software: 10 Top Tools for 2026," September 10, 2026
- The Paypers — "Ramp, AWS sign deal for AI-enabled finance platform," September 7, 2026
- Ramp — "Procurement Case Studies: What 8 Teams Did to See Impact," August 31, 2026
- Financial IT — "Ramp Announces Strategic Collaboration Agreement with AWS," September 7, 2026
- Ramp — "Purchase Order Automation: What It Is & How It Works," September 18, 2026
- RobotWale News — "Figure 02 Humanoid Robot Expands to Global Logistics Pilots with India Market Entry," September 17, 2026
- Financial IT — "Ramp Announces Strategic Collaboration Agreement with AWS to Provide AI-Enabled Finance Operations," September 7, 2026
- Urdupure — "Humanoid Robots in Factories 2026: What Is Actually Real," August 30, 2026
- Complete AI Training — "Ramp and AWS sign strategic deal to bring AI-driven finance," September 5, 2026
- RobotWale — "Humanoids in Logistics: Verified Deployments and Ground Truth," September 5, 2026
- Embodied AI Insights — "Figure AI: Inside the Fastest-Rising Humanoid Robotics Company," September 2, 2026
- Embodied AI Insights — "Figure 03: A Production-Ready Body Still Proving Its Autonomy," August 23, 2026
- GII Research — "Humanoid Robot Industry Progress and Commercialization Research Report, 2026," September 4, 2026
- Forbes Advisor — "The Intelligent Finance Era Is Here: What To Know About Brex's Corporate Card and AI-Native Spend Management," September 4, 2026
- Wiz — "What Is the A2A Protocol? Components, Uses, Security," September 10, 2026
- Upwind — "Upwind vs. Wiz: Pricing, Runtime, and AI Security," September 17, 2026
- A10 Networks — "TrojAI and Wiz: Closing the Loop on AI Security," September 8, 2026
- EPAM — "EPAM Partners with Wiz to Help Enterprises Reduce Cloud Risk," August 24, 2026
- Truto. one — "How to Connect AI Agents to Brex Expense Data via API," September 1, 2026
- PeerSpot — "Netskope Public Cloud Security vs Wiz (2026)," September 11, 2026
- FinChannel — "Software, AI Startups and Cybersecurity: 2026 Surveys Show a Market Moving From Adoption to Accountability," September 10, 2026
- AI Tools Atlas — "Brex for Enterprise: Guide [2026]," September 5, 2026
- Mordor Intelligence — "Enterprise AI Market - Share, Trends & Size 2025 - 2031," September 7, 2018
- AI Central Resources — "Brex vs Airbase (by Paylocity) in 2026: Features, Procurement," September 17, 2026
THE CODEW · STARTUP INTELLIGENCE
Editorial Note
This month's special identifies five companies shaping the next phase of AI-powered enterprise technology — from Wiz and Together AI to Harvey, Figure AI, and AI-driven fintech automation. Each represents a different layer of the emerging enterprise AI stack: security, infrastructure, vertical AI, physical AI, and financial operations.
This coverage is based on public disclosures and reporting cited above and should be read in the context of those sources and their reporting period.
Reviewed by Erwin Castro
on
Tuesday, September 22, 2026
Rating:
