Skip to content

Have something in mind?

Let’s talk

Frontend Engineering

Product Engineering

Swift Agents

Swift Agents is a B2B SaaS platform that helps businesses automate customer conversations using AI-powered support agents.

As a Frontend Engineer at Serendpt AI, I contributed to the platform's product development, proposing and building its Forms feature, developing agent knowledge-base workflows, supporting automated website content refresh, and resolving a production chatbot integration issue.

My work focused on identifying practical product gaps, delivering maintainable frontend features, and improving the experience of businesses integrating Swift Agents into their operations.

Live preview(opens in a new tab)

Project details

Project type
B2B SaaS — AI Customer Support Platform
Role
Frontend Engineer
Engagement
Part-time Employment (Serendpt AI)
Period
April 2026 – Present
Status
Live — Ongoing Development
Technologies
  • Next.js
  • React
  • TypeScript
  • Tailwind CSS
  • TanStack Query
  • React Hook Form
  • Zod
  • REST APIs

Context

Project context.

Swift Agents enables businesses to deploy AI support agents that respond to customer enquiries using information from their websites, documentation, and internal knowledge sources.

Beyond automated conversations, businesses also need ways to collect structured customer information, maintain accurate business knowledge, and reliably integrate customer-facing tools into their websites.

Working within the frontend team, I contributed to these areas through new feature development and improvements to existing functionality.

Problem space

  • Customer data collection was disconnected from support workflows.

    Businesses could use Swift Agents to automate customer conversations, but collecting structured information through forms still required separate solutions. This created an opportunity to bring customer submissions into the same platform.

  • AI agents needed access to current business information.

    Business websites frequently change. New pages, updated services, and navigation changes can make previously collected information outdated, creating a need for more manageable knowledge sources and content refresh workflows.

  • Embedded chatbot reliability directly affected customers.

    Businesses depended on the chatbot integration working correctly on their websites. An integration failure could prevent visitors from interacting with their deployed support agent.

Target users

  • Businesses deploying AI customer support on their websites.
  • Administrators responsible for managing forms, customer submissions, and agent knowledge.
  • Customer support teams using Swift Agents to handle enquiries.
  • Website visitors interacting with embedded forms and chatbots.

Contribution

My contribution.

I worked as a Frontend Engineer responsible for implementing assigned product features, identifying improvement opportunities, integrating frontend workflows with existing APIs, and debugging production issues.

My contributions covered four main areas.

  1. Proposed and Built Swift Agents Forms

    I identified an opportunity to extend Swift Agents beyond customer conversations by allowing businesses to collect structured information through forms.

    Rather than requiring businesses to rely entirely on external form tools, I proposed a native Forms feature and implemented its frontend experience.

    My work included:

    • Building a two-step form creation interface.
    • Supporting hosted and website-embeddable forms.
    • Developing form management and customer submission interfaces.
    • Implementing submission status management, including read and unread states.
    • Integrating creation, retrieval, updates, and deletion workflows with existing APIs.
    • Writing documentation to help users integrate forms into their websites.

    The feature allowed businesses to manage form submissions alongside their existing customer support operations.

  2. Developed Agent Knowledge-Base Workflows

    I built frontend workflows for managing the business information used by Swift Agents' AI support agents.

    This included interfaces for accessing and managing searchable knowledge sources, helping businesses maintain the information their agents rely on when answering customer questions.

    The work supported a more manageable knowledge workflow as businesses updated their products, services, and support information.

  3. Implemented Website Content Refresh Workflows

    I worked on scheduled and on-demand website traversal functionality, internally referred to as strolls.

    These workflows allow businesses to refresh website content used by their support agents without manually repeating the entire information collection process.

    The implementation focused on making content refresh accessible through the product interface and supporting changes to customer websites over time.

  4. Resolved a Production Chatbot Integration Issue

    I investigated a customer-facing issue where the chatbot integration script downloaded successfully but the widget failed to appear on the website.

    After tracing the initialization flow, I identified an obsolete auto-mount condition left behind after an earlier feature was removed.

    I corrected the outdated logic and restored the expected chatbot initialization behaviour.

    This was a production debugging task requiring an understanding of existing application behaviour, rather than simply implementing a new interface.

Scope of ownership

Swift Agents represents the kind of product engineering work I enjoy: understanding how a platform is used, identifying missing capabilities, building maintainable solutions, and resolving problems that affect real customers.

The Forms feature is particularly meaningful because I contributed beyond implementation. I identified a product opportunity, proposed a solution, and delivered the frontend feature through release.

Alongside the knowledge-base and integration work, it demonstrates my experience contributing to a production SaaS platform where product decisions, frontend architecture, and reliability all matter.

Implementation

Technical breakdown.

The frontend was developed using React, Next.js, and TypeScript, with TanStack Query supporting API-driven application state.

My implementation work focused on component-based interfaces, predictable data flows, organization-aware state management, and integration with existing backend services.

  1. Form Creation and Submission Management

    • React
    • Next.js
    • TypeScript
    • React Hook Form
    • Zod
    • TanStack Query

    The Forms feature required multiple connected workflows: creating forms, configuring them, making them available to customers, and managing submitted information.

    I implemented the form creation experience and administrative interfaces, connecting them to the existing API layer.

    Key implementation areas included:

    • Multi-step form creation and editing interfaces.

    • Form input handling and validation.

    • Hosted and embeddable form experiences.

    • Submission retrieval, status updates, and management.

    • API mutations and UI state synchronization.

  2. Organization-Scoped Data Management

    • TanStack Query
    • TypeScript
    • REST APIs

    Because Swift Agents serves multiple businesses, frontend data handling needed to respect the active company context.

    I used organization-scoped TanStack Query keys and appropriate cache invalidation strategies to ensure that form and submission interfaces reflected the correct company and updated after mutations.

    This helped prevent stale or incorrectly reused client-side data when working across company contexts.

  3. Knowledge Management and Website Traversal

    • React
    • TypeScript
    • API integrations
    • TanStack Query

    The knowledge-base and website traversal features required frontend interfaces for managing information used by AI support agents.

    I connected these workflows to existing services, supporting searchable business information and scheduled or manually initiated content refresh operations.

  4. Production Integration Debugging

    For the chatbot issue, I traced the failure through the integration script's initialization path.

    The script was loading, but an outdated conditional prevented the widget from mounting.

    Correcting that logic restored the intended behaviour without requiring businesses to change how they embedded the integration.

Outcomes

What the work delivered.

My contributions expanded Swift Agents' customer-facing capabilities, improved management of business knowledge, and addressed a production integration problem.

  1. Outcome

    Shipped a New Product Capability

    The Forms feature was released, allowing businesses to create forms, embed them on websites, and manage customer submissions within Swift Agents.

    The feature originated from a product opportunity I identified and proposed, rather than an existing implementation brief.

  2. Outcome

    Smooth Initial Forms Integration

    Following the Forms release, the team reported no support tickets relating to form integration.

    Measured impact

    Reported form-integration support tickets after launch
    0

    This was a team-observed support outcome, not a measurement of overall feature adoption or lifetime reliability.

  3. Deliverable

    Improved Knowledge Management

    Delivered knowledge-base and website content refresh workflows that help businesses maintain the information available to their AI support agents.

    These features support ongoing knowledge maintenance as customer websites change.

  4. Outcome

    Restored Customer-Facing Chatbot Functionality

    Identified and fixed an obsolete initialization condition that prevented the chatbot widget from mounting despite its script loading successfully.

    The fix restored expected integration behaviour for the affected customer-facing workflow.