Pragathesh

Lead product designer · AI for commercial banking

A tailored suit fits differently.

Measurements first. Then the pattern. Then another fitting.

You notice it the moment you put it on.

I design AI products the same way: around the people who use them, until the fit is right.

Pragathesh Ravi

Design that fits the people who use it.

7+

Years

Enterprise products and design systems.

50+

Products

Products and platform features shipped.

  • Design systems

    tokens, docs, and contribution

  • Prototypes

    in working code

  • Accessibility

    built in, not bolted on

  • Complex workflows

    in operational software

  • AI decisions

    why the model chose this

  • Research

    through to launch

Experience with

Worked with Esko, Datamatics, and Q2. Logos are trademarks of their respective owners and indicate employment history only.

Bengaluru, India

Expertise

AI Product Design

How I design AI in banking products.

  • Conversational UI

    I design copilot conversations that carry task context so bankers get answers without leaving their workflow.

    68% weekly adoption among eligible users.

  • Agentic Design

    I define what an AI agent can do on its own, when it asks for approval, and how it hands work to a person.

    5% of sessions end in human escalation. The same model applies to back office teams.

  • Predictive Design

    I design recommendations that appear at the point of decision and stay editable by the person accountable for it.

    Recommendations are overridden 12% of the time, and loan processing time dropped 20% over six months.

  • Ambient AI

    I place AI guidance inside existing screens so it supports the task without opening a separate tool.

    Loan officers complete tasks 25% faster.

  • Generative AI Design

    I shape how generated content is drafted, reviewed, and corrected before it reaches a customer or a record.

    8% of outputs are edited by the user before submission.

  • Systems Thinking

    I build interaction patterns that hold across products, teams, and institutions.

    Workflow time dropped 56% with 100% adoption, and patterns run at 150+ financial institutions.

  • Responsible AI

    I surface model uncertainty at the moment of action so users apply the right level of trust.

    Confidence shows only when it is low. 60% of those warnings lead users to seek a second opinion, and 74% of users return within a week after a poor recommendation.