Skip to Content
0%

Meet Bridget: The AI Agent that Bridges Your Systems and Gets Work Done

Three panels show Bridget working in SAP. "Learns in an hour" lists SAP tasks she has learned, such as Create Vendor Master, and their fields. "Executes deterministically" shows her creating a sample supplier, Acme Inc., in seven steps. "Heals automatically" shows the SAP screen switching to German, Bridget relearning the renamed fields, and every value verified after the run. Below, under "Any browser-based system," are the logos of SAP, Oracle Fusion, Workday, Coupa, Salesforce, ServiceNow, and more.
Bridget learns a browser-based application in about an hour, runs each task the same way every time, and repairs the automation herself when the interface changes.

Bridget connects your systems without code, then does the manual work inside them. Generally available in Regrello Nov. 18 and expected in MuleSoft in December.

Picture this scenario: A supplier is approved, the documents are complete, and procurement is ready to place the order. But before that can happen, a person must log into the supplier-management system, work through dozens of screens, and manually enter information the business should already know.

At many companies, the order waits and the work slips. Cycle times stretch, operating costs rise, and contracts and revenue are put at risk.

For decades, IT has gone to great lengths to integrate enterprise applications and automate approvals. But critical work still happens manually in the core systems that run the business — step by step, screen by screen, click by click. This problem is common in the middle and back office where work often moves at the speed of human-first processes. In finance and supply chain, for example, suppliers need to be onboarded, orders processed, and records kept current across systems.

The final step in these processes is usually a person logging in and entering the data by hand. This tedious, manual drag is known as the last mile of a business process.

AI agents hit the same barrier when they don’t have access to a system. When an agent can’t complete the next step in an application, the task goes back to a person.

Meet Bridget, your headless systems integration agent. She quickly connects systems without code, then does the manual work inside your core applications herself. There’s no long, costly integration project to add to the IT backlog. Business teams can improve their own processes and move faster, while IT maintains oversight. 

With Bridget generally available in Regrello on Nov. 18, companies can now turn human-first processes into AI-first operations. Regrello already helps businesses orchestrate complex work across teams and systems. Now, Bridget closes the last mile by doing the manual steps inside those systems, saving time and money and keeping business moving.

Bridget is also expected to be generally available in MuleSoft in December.

Watch Bridget learn, execute, and heal in this demo video:

How Bridget connects systems and executes work

Start by giving Bridget access to a sandbox with a login. In about an hour, she uses AI to explore the application in a browser, learn the processes and tasks, and build a library of governed, reusable MCP tools that agents use to automate each action.

Once your team approves her work, Bridget locks in the code, so agents run each task deterministically, the same way every time, for accuracy, consistency, and predictability.

When the application’s interface changes, Bridget uses AI to diagnose the problem and self-heal, repairing the automation so work gets back up and running quickly.

Bridget runs in your environment using dedicated service accounts, so you stay in control. She also keeps a record of every action she takes and its result, so you always know what happened.

Comparing Bridget with alternatives

With Bridget, business teams don’t have to wait months for IT or consultants to build custom integrations. They don’t have to rely on brittle RPA, which may break when an application’s interface changes. 

Newer AI agents use computer vision to navigate applications on their own, figuring out each step as they go. That makes them unpredictable and expensive to run on critical work. To measure the difference, we performed 250 hours of benchmark testing to compare Bridget against a frontier model across easy, medium, and hard workflows in enterprise applications.

Accuracy: Bridget achieved 100% accuracy across all three tested workflows. The frontier model, by contrast, struggled more as complexity grew. On the hardest workflow, it even sent emails to the wrong supplier.

Cost and speed: Bridget also completed the workflows faster and cheaper, with zero model cost at run time.

Bridget combines the best of both worlds, using AI to explore and learn and software to run each task deterministically. That means you can finally bring AI into your critical operational processes with the reliability and predictability your business requires.

How the supplier handoff looks with Bridget

Let’s revisit the scenario from the beginning of this post.

When the supplier was approved, the order stalled while it waited for someone to manually enter information the business already had.

With Bridget, here’s what that handoff looks like now.

First, a Regrello blueprint sends the approved supplier information to Bridget. Next, Bridget creates the supplier in the supplier-management system and verifies the result. Finally, Bridget returns that result so the order can move forward, all with a complete audit trail. 

The same approach applies to entering approved orders, updating shipping details, maintaining records across systems, and more. Regrello orchestrates the broader process, and Bridget handles the steps inside each application.

Customers are cutting manual work from weeks to days

Freeport-McMoRan, a leading international metals company, uses Bridget to automate previously manual work across its supplier-onboarding process. Bridget quickly built five workflows in a sandbox over a weekend, followed by testing and a production rollout. As a result, supplier-onboarding time dropped from weeks to days, and the system work has been automated for more than 400 suppliers to date.

According to Freeport-McMoRan: “In three months, our business users have built dozens of workflows with Bridget across our enterprise systems. In material deduplication alone, Bridget is helping us pursue significant savings and was key to success by automating more than 20 distinct transactions in SAP S4. We can now tackle opportunities that were previously impractical at scale.”

Global semiconductor manufacturer Microchip Technology is another Regrello customer piloting Bridget to automate previously manual work inside core systems like ERP.

“Bridget enables us to achieve rapid automation of browser-based business processes, providing a faster alternative in the absence of APIs,” said Theresa Mouton, Senior Director of IT Enterprise Applications at Microchip Technology. “For example, we are doing a pilot with Bridget to automate vendor creation in our ERP system, transforming a manual process into an automated workflow.”

Here at Salesforce, we’re Customer Zero of Regrello, with more than 60 internal use cases. For example, our finance team is using Bridget to execute forecast adjustments in its budgeting application. Previously, the team had to manually enter about 70 of these adjustments every quarter. Bridget now handles the entries while the team simply reviews the results.

Choose your first task for Bridget

Give Bridget the work your team is tired of doing by hand. Good first tasks are repetitive, follow clear business rules, and run in a browser application with a sandbox.

Ready to put Bridget to work? Learn more about Bridget and Regrello and see how Salesforce runs on Regrello in our AI-First Operations webinar.

See Salesforce run on Regrello

Salesforce is Customer Zero for Regrello and uses it across its own back-office work. The AI-First Operations webinar walks through real use cases, a live demo, and how our team is scaling AI-first operations.

Get the latest articles in your inbox.

Get the latest articles in your inbox.