OpenAI Presence Review 2026: Enterprise AI Agent Platform — Deep Dive Analysis

Your enterprise has been waiting for AI agents that actually work in production. Not demos. Not proof-of-concept chatbots that collapse under real customer traffic. Agents that answer calls, resolve billing issues, verify identities, follow company policy, and escalate when they should — all while staying within the guardrails your compliance team demands.

OpenAI just shipped the product it built for that job.

On July 22, 2026, OpenAI launched Presence — a full-stack enterprise platform for deploying AI-powered voice and chat agents across customer support, outbound sales, IT services, and HR workflows. It is not another API. It is not a model endpoint. Presence is a deployed product, led by OpenAI Forward Deployed Engineers (FDEs), backed by the newly formed $10B OpenAI Deployment Company, and already running 75% of OpenAI’s own English-language phone support without human intervention.

This is the most important enterprise AI product launch of 2026. Here is exactly what it does, what it costs, who it fits, and — critically — where it falls short.

---

Bottom Line Up Front

OpenAI Presence is the most ambitious enterprise AI agent platform on the market, but it is not for everyone. If your organization handles high-volume customer support, insurance claims, billing operations, or IT service requests — and you have the budget and tolerance for a high-touch deployment model — Presence delivers production-grade AI agents with governance, continuous improvement, and a proven track record (OpenAI uses it for its own support line). It resolves 75% of inbound calls autonomously and reduced human handoffs by 15 percentage points in 10 days on OpenAI’s own deployment.

The catch: Presence is not self-service. It is not available via API. It requires Forward Deployed Engineers, custom scoping, and enterprise negotiations. There is no published pricing. It competes directly with Salesforce Agentforce ($2/conversation), ServiceNow AI Agents (bundled into ITSM tiers at $100-150/user/mo), and Zendesk AI ($1/automated resolution) — but with a fundamentally different go-to-market model.

Winner for most enterprises: Wait for broader availability or evaluate Salesforce Agentforce (if you are on Salesforce) or ServiceNow AI Agents (if you run IT workflows). Presence is the right choice only if you want OpenAI-led deployment, need a greenfield agent platform, and have the resources for a high-touch engagement.

---

Quick Comparison: Enterprise AI Agent Platforms

	OpenAI Presence	Salesforce Agentforce	ServiceNow AI Agents	Zendesk AI
Best for	High-volume voice/chat support with governance	CRM-embedded sales/service agents	IT and HR workflow automation	Customer service teams
Deployment model	FDE-led, custom scoping	Self-service + partners	Self-service + partners	Self-service + partners
Pricing	Custom enterprise (not published)	$2/conversation or $125/user/mo	Bundled in ITSM ($100-150/user/mo)	$1/automated resolution
Voice support	✅ Real-time (GPT-Live)	❌ Limited	❌ No native voice	✅ Via integrations
Self-healing	✅ Codex-powered improvement loop	❌ Manual updates	❌ Manual updates	✅ Forethought AI
Self-service	❌ Not available	✅ Available	✅ Available	✅ Available
Best for regulated industries	✅ High (FDE-led compliance)	⚠️ Moderate	✅ High (FedRAMP)	⚠️ Moderate
Go-to-market	OpenAI FDEs + systems integrators	Self-serve + partner ecosystem	Partner-led + direct sales	Self-serve + partner ecosystem

---

What Is OpenAI Presence?

Presence is OpenAI’s answer to the hardest problem in enterprise AI: getting agents to behave reliably in production, at scale, under real-world conditions, with genuine governance. It is a deployed product — not a model, not an API — that packages everything needed to run AI agents in an enterprise environment:

• Policies and SOPs — Define exactly what the agent can and cannot do
• Guardrails — Automatic intervention when interactions go outside boundaries
• Approved actions — Grant specific system access (billing, refunds, account changes)
• Simulations and evaluation tools — Test agents against common and edge-case scenarios before launch
• Codex-powered improvement — Continuous learning from production sessions and escalations

The platform is available today for real-time voice and chat experiences. Each deployment starts with a specific job — resolving billing issues, supporting insurance claims, handling employee IT requests — and the agent receives only the knowledge and system access required for that task.

---

How Presence Works

The Deployment Model

Presence flips the traditional SaaS playbook. Instead of “sign up, configure, go live,” OpenAI sends Forward Deployed Engineers (FDEs) to work inside your organization. These engineers:

1. Identify high-value workflows — Customer support, billing, IT ticketing, HR requests
2. Connect necessary systems — CRMs, ticketing platforms, knowledge bases, billing systems
3. Establish permissions and policies — What the agent can do, what needs approval, when a human takes over
4. Test the agent — Against common requests, edge cases, and high-risk scenarios
5. Deploy to production — With ongoing monitoring and improvement

This is the Palantir playbook applied to AI agents, and it is deliberate. OpenAI is not selling software; it is selling outcomes.

The Continuous Improvement Loop

This is Presence’s genuinely unique feature. After launch, the agent does not stagnate. Codex — OpenAI’s coding agent — continuously reviews production sessions, escalations, and quality signals. It identifies patterns where the agent is underperforming and proposes behavioral updates. Human teams test each proposed change against the production version before approving a controlled rollout.

The result: on OpenAI’s own support line, handoffs to humans dropped by 15 percentage points in just 10 days of this improvement loop running.

Voice and Chat Capabilities

Presence uses GPT-Live for real-time voice conversations. The agent can:

• Understand open-ended requests in natural language
• Verify caller identity when needed
• Use relevant account context
• Take approved actions (billing resolution, account changes, refunds)
• Escalate to human agents when required

---

Deep Dive: Key Features

1. Policy Engine and Guardrails

The governance layer is where Presence separates itself from building agents on raw APIs. You define policies in plain language — “refunds over $500 require manager approval” — and the guardrails enforce them automatically. The system checks every action against defined permissions before execution.

What we liked: The policy layer is genuinely production-grade. It is not a “suggestion” system — it enforces boundaries. Companies in regulated industries (insurance, banking, healthcare) will find this compelling.

What we didn’t: The policy configuration requires OpenAI FDE involvement. You cannot self-serve this setup. If your compliance team wants to iterate on policies daily, you are dependent on OpenAI’s deployment schedule.

2. Simulation and Evaluation Framework

Before any agent touches real customers, Presence runs it through a battery of simulations covering common requests, edge cases, and high-risk scenarios. Graders evaluate whether the agent:

• Reached the intended outcome
• Followed company policy
• Used tools correctly
• Escalated when appropriate

What we liked: This is the feature most enterprise AI deployments skip, and it is why so many fail. Presence makes simulation a first-class part of the deployment process. The grader framework is transparent — you can see exactly why an agent passed or failed a scenario.

What we didn’t: The simulation library is built per-deployment. There is no shared community library of scenarios across industries. Each customer essentially starts from scratch on their industry-specific edge cases.

3. Codex-Powered Continuous Improvement

The most differentiated feature of Presence. Production sessions, escalations, and quality signals flow back to Codex, which investigates patterns and proposes updates. Teams can A/B test proposed changes against the production version before rolling them out.

What we liked: This closes the loop that most agent platforms leave open. Your agent gets measurably better over time without requiring you to rebuild it. The 15-percentage-point handoff reduction in 10 days on OpenAI’s own line is evidence this works in practice.

What we didn’t: The improvement loop is dependent on Codex, which is itself an AI. If Codex misdiagnoses a pattern or proposes a change that introduces new issues, the review process is your safety net — but that review process is manual.

4. Multi-Channel Support

Presence currently supports real-time voice and chat. OpenAI’s marketing materials mention broader ambitions (email, SMS, messaging apps), but the company has not confirmed email support at launch.

What we liked: The unified framework — same policies, same guardrails, same evaluation tools — works across channels. You define governance once and apply it everywhere.

What we didn’t: Voice-only and chat-only at launch. If your operation runs on email, SMS, or WhatsApp, you will need to wait for expansion or use a separate platform.

---

Pricing Breakdown

OpenAI has not published pricing for Presence. This is the single biggest question mark in the entire product launch.

Here is what we know:

• Deployments are scoped individually based on each customer’s use case
• Pricing includes the FDE engagement, platform subscription, and usage
• The company has said “broader pricing details to come as availability expands”

For context, here is what competitors charge:

Platform	Pricing Model	Typical Cost
Salesforce Agentforce	Per-conversation or per-user	$2/conversation or $125/user/mo
ServiceNow AI Agents	Bundled in ITSM tiers	$100-150/user/mo (with 20-40% renewal uplift)
Zendesk AI	Per-automated-resolution	$1/resolution (Advanced tier, custom pricing)
Microsoft Copilot Studio	Credit-based	$200/mo for 25K credits ($0.008/credit)
Intercom Fin	Per-resolution	$0.99/resolution

Our estimate: Given the FDE-led deployment model and OpenAI’s investment in the Deployment Company ($10B valuation), we expect Presence to cost significantly more than self-service alternatives — likely in the range of $10,000-$50,000+/month for the platform, plus FDE engagement fees ($2,000-5,000/day per engineer). This is not a product for small or mid-market businesses.

---

Who Is Presence For?

Right fit

• Large enterprises ($500M+ revenue) with high-volume customer support, billing, or claims operations
• Regulated industries (insurance, banking, healthcare) where governance and audit trails matter more than price
• Organizations already using OpenAI who want a single-vendor agent platform
• Companies willing to adopt a high-touch, FDE-led deployment process

Wrong fit

• SMBs and mid-market companies — Presence is not available as self-service and will not be cost-effective at smaller scales
• Teams that need an agent platform this week — The FDE-led deployment means months, not days
• Companies on a strict per-seat or per-conversation budget — Presence pricing is opaque and likely premium
• Organizations that want model flexibility — Presence uses OpenAI models for the core agent (though it can connect third-party models for guardrails and tools)

---

The Competitive Landscape

vs Salesforce Agentforce

Salesforce Agentforce ($2/conversation, $125/user/mo) is Presence’s most direct competitor on customer service use cases. Agentforce is available now, has published pricing, and works inside the Salesforce ecosystem. Presence wins on governance depth, the Codex improvement loop, and voice-native capabilities. Agentforce wins on availability, self-service setup, and CRM-native data.

If you are already on Salesforce: Start with Agentforce. Presence is interesting but unproven outside OpenAI’s own support line.

vs ServiceNow AI Agents

ServiceNow AI Agents are bundled into ITSM tiers ($100-150/user/mo) and focus on IT and HR workflow automation. ServiceNow wins on IT service management depth, FedRAMP compliance, and enterprise workflow orchestration. Presence wins on voice support, continuous improvement, and customer-facing agent use cases.

If your primary need is IT/HR automation: ServiceNow is the clear choice. Presence is for customer-facing agents.

vs Zendesk AI

Zendesk AI ($1/resolution, Advanced tier) is the most accessible enterprise agent platform. It has transparent pricing, a self-service setup, and pre-trained AI that works out of the box. Presence wins on governance depth and the continuous improvement loop. Zendesk wins on ease of setup, pricing transparency, and breadth of language support (80+ languages vs Presence’s undisclosed count).

If you want AI working this month: Zendesk AI is the better choice. Presence requires a multi-month deployment.

---

The Verdict

Criteria	Score (out of 10)
Governance & guardrails	9/10
Voice capabilities	9/10
Continuous improvement	9/10
Ease of deployment	3/10
Pricing transparency	2/10
Self-service availability	1/10
Production evidence	8/10
Ecosystem breadth	5/10
Overall	6.5/10

OpenAI Presence is a genuinely impressive product with a genuine go-to-market problem. The governance layer, Codex improvement loop, and production evidence (75% autonomous resolution, 15-point handoff reduction in 10 days) are best-in-class. The deployment model — FDE-led, custom scoped, enterprise negotiated — is the same model that made Palantir a $50B company and Salesforce Agentforce a $1.2B ARR business in 18 months.

But Presence is not ready for the broad enterprise market. Without published pricing, self-service availability, or clear SLAs, it remains a product for early adopters with deep pockets and high tolerance for vendor-led deployment.

For most enterprises: Evaluate Salesforce Agentforce or Zendesk AI if you need agents now. Book a Presence demo if you want the most advanced governance and improvement loop in the market and have the resources for a high-touch engagement.

For the Fortune 500 with a dedicated AI budget: Presence is worth the conversation. The Codex improvement loop alone could justify the premium over platforms that require manual updates.

---

Frequently Asked Questions

What is OpenAI Presence?

OpenAI Presence is an enterprise AI agent platform launched July 22, 2026. It enables companies to deploy AI-powered voice and chat agents for customer support, sales, IT, and HR workflows, with built-in governance, guardrails, and a Codex-powered continuous improvement loop.

How much does OpenAI Presence cost?

OpenAI has not published pricing for Presence. Deployments are scoped individually based on each customer’s use case and implementation needs. Pricing includes the platform subscription and Forward Deployed Engineer engagement. Broader pricing details are expected as availability expands.

Is OpenAI Presence available as a self-service product?

No. Presence is available through a limited general availability program, with deployments led by OpenAI Forward Deployed Engineers and select global systems integrators. It is not available as a self-serve product.

Does Presence use only OpenAI models?

Presence uses OpenAI models for the core agent logic. It can connect third-party models and services through APIs for guardrails, tools, and other components. The company has not confirmed which specific models power the core agent.

How does Presence compare to building agents on the OpenAI API?

Building agents directly on the OpenAI API gives you full control but requires you to build your own governance, guardrails, simulation testing, monitoring, and continuous improvement infrastructure. Presence packages all of these into a single deployed platform with FDE support. For most enterprises, the total cost of building and maintaining these systems internally exceeds Presence’s likely pricing.

What use cases does Presence support?

Presence currently supports real-time voice and chat for customer support, outbound sales development, procurement, IT services, and HR functions. Specific examples include billing resolution, insurance claims handling, account changes, refund processing, and employee IT service requests.

Who are Presence’s early customers?

BBVA Mexico, SoftBank Corp., and Retail Insurance Australia (part of IAG) are the publicly named early customers. OpenAI also uses Presence to run its own English-language phone support line.

Does Presence meet enterprise compliance requirements?

Presence is designed for regulated industries with SOC 2 compliance, encryption at rest and in transit, and the FDE-led deployment model ensures compliance requirements are addressed during implementation. OpenAI has not published specific compliance certifications or SLAs.

---

Disclosure: Some links in this post are affiliate links. If you click through and take action, we may earn a small commission at no additional cost to you. We only recommend products we have thoroughly researched and believe deliver genuine value.