top of page

Hi! I'm Nishka

Pink Poppy Flowers

  • Audited existing docs for tone, structure, and accessibility gaps

  • Created the Instructions Doc and standardized overview template

  • Built a prompting system for consistent output

  • Tested on components (Tab Bar, Progress, Step Bar) and refined iteratively

  • Partnered with engineering on repo structure and automation alignment

  • Validated quality by rewriting and generating pages for accuracy and consistency

  • Automation: Auto-generate MRs with updated docs and visual assets

  • Workflow Expansion: Generate full workflow blueprints and detect pattern gaps

  • Quality: Run automated audits for tone, structure, and accessibility compliance

  • Integrations: Connect to Figma and the DS site for one-click doc generation and publishing

Pink Poppy Flowers
Pink Poppy Flowers

Accessibility

Supports longer labels and complex flows, improves clarity and scannability, replaces inconsistent custom steppers, reinforces ‘one step = one task,’ reduces cognitive load with a top-to-bottom layout, and creates a unified, future-proof workflow pattern.”

Splunk’s docs pipeline was slow and inconsistent, so I built an AI copilot that generates fully formatted, on-brand component pages and audits existing content for clarity and tone.

An AI Copilot built around structure, constraints, and reusable patterns​

Instead of pulling raw metadata, the copilot uses a curated foundation:

  • A detailed Instructions Document defining tone, structure, accessibility expectations, usage guidance, and system best practices.

  • A Prompting Architecture that chains these rules into consistent generation steps.

  • A Skeleton Overview Template that standardizes the layout for all component pages.

  • References to existing Splunk UI documentation to ensure style and voice alignment.

​​

The copilot agent can:

  • Generate new overview pages aligned to Splunk UI design systems voice & structure

  • Produce consistent Dos & Don’ts, usage guidance, accessibility notes

  • Review and rewrite existing pages for tone, clarity, and completeness

  • Enforce writing patterns, accessibility rules, and narrative consistency​

AI Documentation Copilot

Accelerating design system documentation with AI-powered automation

Splunk UI's documentation pipeline became a major bottleneck - each component page demanded heavy writing, inconsistent voice, and long review loops. I built an AI copilot that uses structured instructions, a prompting framework, and skeleton templates to generate fully formatted, Splunk-aligned pages and audit existing content for clarity, tone, and completeness.

Overview

The problem

Documentation was slow, inconsistent, and dependent on individual writing styles

  • Contributors wrote pages with vastly different tones, structures, and levels of detail.

  • Pages required multiple review cycles to fix inconsistencies.

  • Documentation became the slowest part of component release.
     

No shared framework for what “good documentation” looked like

  • Contributors lacked a unified standard for tone, depth, structure, accessibility notes, or narrative framing.

  • Even similar components had wildly different writing quality.
     

SUI was scaling too fast for manual doc production

  • As SUI shifted toward workflows and complex patterns, writing documentation by hand didn’t scale.
     

We needed a fast, repeatable, high-quality way to generate and maintain documentation—one that preserved voice, structure, and accessibility accuracy.​​​​​​​​​​​​​

The solution

An AI Copilot built around structure, constraints, and reusable patterns​

Instead of pulling raw metadata, the copilot uses a curated foundation:

  • A detailed Instructions Document defining tone, structure, accessibility expectations, usage guidance, and system best practices.

  • A Prompting Architecture that chains these rules into consistent generation steps.

  • A Skeleton Overview Template that standardizes the layout for all component pages.

  • References to existing Splunk UI documentation to ensure style and voice alignment.

​​

The copilot agent can:

  • Generate new overview pages aligned to Splunk UI design systems voice & structure

  • Produce consistent Dos & Don’ts, usage guidance, accessibility notes

  • Review and rewrite existing pages for tone, clarity, and completeness

  • Enforce writing patterns, accessibility rules, and narrative consistency


The result is a documentation engine that reduces manual writing while increasing system reliability.

Process

Audited existing documentation for tone, structural gaps, and accessibility inconsistencies.

​

Created the Instructions Doc to unify best practices, voice, narrative patterns, and accessibility rules.

​

Designed the overview page template to standardize hierarchy and content structure.

Foundation & Standards

Built a chain-based prompting system to enforce writing rules and generate predictable outputs.

​

Tested generation on existing components (Tab Bar, Progress, Step Bar).

​

Refined tone, voice, and structure constraints through iterative testing.

AI System Design 

Cross-Team Alignment 

Partnered with engineering on repo expectations, asset handling, and future automation opportunities.

​

Evaluated performance by rewriting existing pages and generating new ones to ensure accuracy, consistency, and repeatability.

Scale

The problem

  • Documentation was slow, inconsistent, and required multiple review cycles due to differing writing styles.

  • No shared standards for tone, structure, or accessibility led to uneven quality across components.

  • Manual docs couldn’t keep up with SUI’s rapid growth and increasing complexity.

  • We needed a fast, consistent, scalable way to produce high-quality documentation.

The solution

Discovery - interviews, audits, Design System research, hierarchy mapping.


Alignment - Worked with accessibility, engineering, and Cisco Design System teams.


Delivery -  Built the vertical variant, iterated with engineering, and documented.

The process

Pink Poppy Flowers

Product impact

Before

Pink Poppy Flowers

After

Delivered a standardized vertical Step Bar that replaced ad-hoc versions, unified workflows, and reduced design/engineering debt. Added a clear, accessible heading model, aligned patterns with Magnetic, and strengthened ARIA, DOM, and keyboard standards through close engineering and a11y partnership.

Result

  • Uncontrolled Flexibility

  • Accessibility and interaction problems

  • Alignment without forced parity

  • Technical constraints

Challenges

Internal Tooling

Documentation Standardization

Prompt Engineering

Workflow acceleration with AI

Pink Poppy Flowers
Pink Poppy Flowers
Pink Poppy Flowers
Pink Poppy Flowers

Missing overview page for anchor

Layout + structure inconsistencies

Missing overview page for Anchor component

Layout and structural inconsistency 

I led the design, strategy, and prompting architecture behind the copilot

  • Defined the documentation standards for tone, structure, and accessibility

  • Authored the Instructions Doc that teaches the model how to write for SUI

  • Designed the Prompting Framework that guides generation and review

  • Built the Skeleton Template used to create uniform overview pages

  • Partnered with engineering to validate repo structure, publishing feasibility, and MR considerations

  • Conducted iterative testing with real components to refine accuracy

  • This work combined design-system expertise, writing systems, prompt engineering, and product strategy.

  • Cuts 30+ hours of manual writing effort per week across Splunk UI

  • Unifies tone and structure across all component documentation

  • Shrinks review time thanks to consistently structured, high-quality drafts

  • Improves accessibility standards by enforcing WCAG-compliant writing

  • Reduces design and engineering documentation debt, accelerating system growth

Pink Poppy Flowers
Pink Poppy Flowers

Impact

Pink Poppy Flowers

I led the copilot’s design, strategy, and prompting architecture - defining documentation standards, authoring the Instructions Doc, creating the prompting framework, and building the skeleton template. I partnered with engineering on repo and publishing feasibility and refined the system through iterative testing on real components.

My role

Process

  • Saves 30+ hours of manual writing per week

  • Unifies tone and structure across all docs

  • Shortens review cycles with consistent drafts

  • Enforces WCAG-aligned accessibility standards

  • Sets the stage for workflow-level doc automation

  • Reduces documentation debt across design and engineering

Impact

  • Automation: Auto-generate MRs with updated docs and visual assets.

  • Workflow Expansion: Generate full workflow blueprints and detect pattern gaps.

  • Quality: Run automated audits for tone, structure, and accessibility compliance.

  • Integrations: Connect to Figma and the DS site for one-click doc generation and publishing.

Impact

Home

Work

Resume

Nishka.jiandani@gmail.com
Tel. (415) 610 6284

San Francisco, CA

bottom of page