Precog
PRODUCT DESIGN
WEB APP
AI
Role
UI Redesign, UX Design,
User testing
Impact
Faster Workflows,
Fewer Errors, High User Adoption
Team
1 Designer, 2 Project Managers,
6 Developers , 2 ML Engineer
Overview ✨
What is Precog
Precog is a bot management platform that hosts and manages fnFM Bots. These bots execute the logic built in Pi Design Studio. Precog operates through three critical components:
Processor – the decision-making brain
Connector – linking incoming data to Fabrix metadata
Bot – executing rules and delivering processed outputs
As the existing tool grew in complexity, users struggled with discoverability, configuration clarity, and bot monitoring. My redesign focused on making the system more intuitive, scalable, and efficient for both technical and semi-technical users.
Goal 🚀
Simplify bot creation, configuration , and management.
Provide a visual hierarchy of processors, connectors, and bots.
Enable scalable workflows by making node management effortless.
Improve system feedback for actions, dependencies, and errors.
Outcome
Faster
Workflows
Fewer
Errors
High User
Adoption
Before
After
50% Faster Workflows – Streamlined navigation and simplified interactions cut down unnecessary steps.
Fewer Errors – Clear validations and improved information hierarchy reduced user mistakes.
Higher User Adoption – Modern visuals, consistent patterns, and responsive design increased overall usability and trust.
How It Works
A bot in Precog is an execution unit. To build a bot, you need three building blocks:
Source - Where the input data comes from (e.g., SharePoint folder, Google storage, input form).
Processor - A Processor is basically the logic engine that sits between the Source and the Destination.
Destination - Where the processed/structured output should be stored or sent (e.g., JSON output, database, another folder).
Case Example
Case Example - Automating Speciality Drug Enrollment
A major pharma client needed to digitize thousands of handwritten specialty drug enrollment forms and compare it with database and send it for prior authorization.
Source:
The pipeline starts by connecting to a SharePoint folder link where all the files are stored. PreCog automatically fetches the required documents (Excel, CSV, PDF, etc.) from the link.
Processor:
Once the files are ingested, they are passed through the built-in processors inside PI Design Studio. Here, business logic is applied for example, cleaning data, extracting key values, applying validations, or transforming the format.
Destination:
After processing, the output is converted into structured JSON. This JSON can then be used by downstream applications, APIs, or analytics tools for further workflows.
Key UX Challenges
Problem : Lack of Bot Run Visibility
Users had no way to track what happened after running a bot. If a run failed or partially succeeded, there was no clear breakdown of which step failed, what data was skipped, or how long the execution took. This made debugging time-consuming and reduced trust in the system.
Solution:
Introduced a Bot Run Logs panel that captures:
Step-by-step execution logs with timestamps
Color-coded categories: Info, Debug, Warning, Error
Human-readable summaries (e.g., skipped records)
Aggregated run summary: errors, skips, time, records processed
Export option for audit/compliance
Problem : Fragmented Connector & Processor Creation
While creating a bot, users could not add new connectors or processors within the flow. They had to exit the setup, navigate back to a separate screen to create them, and then return to the bot flow. This broke continuity, caused confusion, and added unnecessary steps — especially frustrating for first-time users.
Solution:
We introduced in-context creation for connectors and processors. At each relevant step, users can now click “Add New”, which opens a side panel. This allows them to configure a connector or processor inline, without leaving the flow. Once added, it’s instantly available in the current step.
Problem : Lack of Context in Toast Messages
Users often received generic toast notifications (e.g., “Action failed” or “Success”), which lacked context about the action, location, or next steps. This caused confusion, reduced trust in system feedback, and forced users to check logs or retry blindly.
Solution:
Introduced context-rich toast notifications with:
Clear action + outcome (e.g., “Data Push Failed – Network timeout while sending records to Patient Data”).
Guidance on next steps (e.g., “Click here to view logs”).
Consistent tone and placement across modules.
Color-coded severity (success, warning, error) for instant recognition.
Designing the Solution
Provides a high-level overview of the entire Precog system, giving quick insights into activity and status at a glance.
Displays the latest bots along with their recent runs and the status of each run for easy tracking.
A side panel to quickly add a source and destination, simplifying the setup process.
Create Processor is where logic from PI Design Studio is connected that sits between the Source and the Destination.
Provides an overview of all schedules with status indicators (Active, Failed), start dates, and completion insights, helping users quickly track automation reliability.
Create Schedule Panel: A side panel that allows users to configure a new bot run by selecting the bot, labeling the schedule, setting start date/time, and defining custom recurrence (daily, weekly, or specific days).
User Testing 🏆
Conducted testing with internal users (designers, data engineers, and business users) to validate the bot creation flow.
Observed how users interacted with Source, Processor, Destination setup and whether they understood the logic without external guidance.
Tested toast notifications, error handling and confirmation prompts during bot creation.
Measured how easily first-time users could complete the flow without documentation.
Made with Love, Framer,
and a lot of Ctrl+Z