Forethought's Browser Agent Takes Action Anywhere a Human Can Click

According to Postman's 2025 State of the API report, 85% of enterprise systems don't have public APIs. Systems like your ERP from 2008, your legacy billing system, the internal admin tools someone built in-house five years ago, your old CRM that still works fine but never got modernized, might run your business, but AI agents can't reach them without an API.

Forethought's Browser Agent allows AI agents to access systems without APIs. An AI agent can log in, click, type, and navigate through any system a human can access through a browser, following the policies you define without an API. It's built for enterprise security and works with your existing Forethought setup across every channel.

“AI has spent the last few years learning to answer questions,” said Sami Ghoche, CEO and Co-Founder of Forethought. “The next era is about action and outcomes. Browser Agent allows AI to securely complete real tasks inside systems that were never built for automation. It helps enterprises make a meaningful impact without changing the systems they already rely on.”

Older systems weren't built for automation, and replacing them is rarely practical. Browser Agent removes one of the biggest blockers to agentic AI adoption in the enterprise, allowing you to automate the customer experience even when your systems can’t.

Most enterprise work lives where AI can't reach

Behind successful agentic AI adoption is the API. Implementations are successful when someone exposes an API endpoint that the AI can call. For example, a refund system and an identity management system each have APIs. If exposed, AI can trigger actions through them to process a refund.

In reality, many vendor systems were never built for this type of automation. According to Postman's research, only 24% of developers design APIs with AI agents in mind, and 31% of organizations are managing multiple API gateways just to keep everything connected. Without this capability, an AI agent could still deflect a meaningful number of customer questions by understanding customer intent, pulling the right policy, and explaining what needs to happen. However, a human agent would still have to do manual work to resolve many issues.

Say you run a pharmacy retail chain. A customer calls asking for a prescription refill, and your AI agent understands the request. It verifies the customer's identity, checks their prescription history, confirms the refill is due, and reviews HIPAA compliance requirements. But your pharmacy management system doesn't have an API, so it can't actually submit the refill request.

Instead, it collects the information and passes the customer to a human agent for fulfillment. It tells the customer, "I've gathered your information and passed your request to a pharmacist who will review it within 24 hours and contact you when it's ready." The customer waits while a pharmacy technician logs in and manually processes the refill. If the request is denied for any reason, like an insurance issue, interaction warning, or refill too early, the customer won't know until the next day. Then they'll have to call back and restart the process.

Forethought Browser Agent closes the API gap

Browser Agent changes that. It acts inside browser interfaces exactly the way a trained support rep would, but consistently, instantly, and at scale. It clicks buttons, fills forms, navigates pages, submits actions, watches for confirmations, and even handles multi-step processes in any system a human can access through a browser.

“Agentic AI isn’t just another technology trend; it’s redefining how businesses operate and build value," said Antoine Nasr, Head of AI at Forethought. “Our mission has always been to make AI trustworthy and impactful for teams solving real problems. What our customers have proven is that when companies design AI around outcomes, they create measurable results.”

Suppose AI can navigate through your systems the way a human agent can. In that case, you need control over what it accesses, when it requires approval, and complete visibility into every action it takes. Browser Agent provides capabilities that give you that control:

• Natural language action builder: Define steps in plain English, no code.
• Omnichannel execution: The same actions work across chat, email, voice, and Slack.
• Human-in-the-loop controls: Require review for high-impact steps.
• Enterprise guardrails: Domain allow-lists, encrypted credentials, simulation mode, and role-based access.
• Complete audit visibility: A step-by-step action log for compliance and QA.

How Browser Agent works in practice

Browser Agent works by following policies written by your team. Each policy specifies which system to open, the actions to take within that system, and how to verify success. When the policy is triggered, Browser Agent logs in, navigates the system, and performs the steps in real time. It follows your defined path precisely, clicks and types like a human would, and reports back once the task is complete.

For example, a subscription company that manages customer accounts through an internal dashboard might receive a chat message that says, “I want to change my plan from Basic to Pro.”

The AI agent identifies the intent (“change account plan”), and the Browser Agent then begins executing the policy the CX team has defined for this task. In the policy editor, the team has described the exact steps Browser Agent should take: log in to Salesforce, find the customer, edit the plan type, save, and confirm.

Once triggered, Browser Agent executes the process just like a human would. It opens the browser interface, navigates through the customer database, searches for the correct user record, clicks into the plan type field, updates it, and confirms that the change succeeded. At the same time, your AI agent communicates progress to the customer.

The customer experiences what feels like a natural, short wait: “Performing your request… this usually takes a minute or two.” Then, “Your account has been upgraded to Pro.” What would usually take a human agent several minutes happens automatically in under one.

The same policy works seamlessly in other channels. During a voice interaction, for instance, Browser Agent can execute the same “change account plan” policy in the background. At the same time, the AI handles the live conversation, including verifying the account, confirming the new plan, and updating the system in real time.

Browser Agent makes every customer interaction actionable, no matter how old or isolated the underlying systems are. It enables you to automate workflows of all types, including data-source connections, trade executions, account updates, password resets, and more.

Unlock the 76% of enterprise work trapped behind browser interfaces

Most enterprise work doesn’t live in clean, API-connected systems. It lives behind browser logins—inside legacy tools, internal dashboards, vendor portals, and compliance-restricted environments.

Now, instead of optimizing around the limits of your tech stack, teams can finally automate based on how work actually gets done. Browser Agent can now access:

• Every legacy system without API access
• Every internal tool your team built in-house
• Every vendor platform that refuses to give you API keys
• Every compliance-restricted system where API access isn’t allowed

The gap between what AI understands and what it can do has just closed. Book a Forethought demo to see how Browser Agent unlocks any system regardless of its API status.