Saraschandra KaranamHuman - AI Interactions Researcher · Enterprise AI
Back to Research
RESEARCH PROGRAM • 2017-26

Complex Enterprise Workflows

How to empower domain experts to orchestrate complex infrastructure using automation or by interacting with an abstracted representation of a system?

Research /EvaluationPurchaseStagingDay 0onboardingDay NconfigurationMonitoringTroubleshootingUpgrades
Representative Visualization

CONSOLIDATED JOURNEY MAP

Journey map showing various stages a typical network or security professional would go through when working with network or security devices

40+Studies
200+Leads
4Platforms
02 · WHY THIS RESEARCH MATTERED

Closing the Translation Gap: Optimizing Automation Ecosystems to Prevent Systemic Failures

Optimizes the interfaces that eliminate the human-to-machine translation gap, allowing domain experts to deploy complex systems without coding bottlenecks. By validating abstractions and simulation guardrails, it directly reduces human error, democratizes advanced technical workflows, and prevents catastrophic real-world failures.

After (Catalyst Center GUI)
Cisco Catalyst Center — Software UpgradeImage UpgradeIntegrity CheckDevice StatusSystem SettingsCatalyst 9300 Switch Stack (3 Switches)Software Version IOS-XE 17.9.3 — Ready for ActivationUpgradeAutomated Verification StepsPlatform cleanup checkPassedPackage staging & integrity (MD5 verified)PassedSoftware package activationPassedPost-upgrade version validationValidating...80% Complete
Before (CLI)
# Preparation
Switch# request platform software package clean switch all
Switch# copy tftp://10.1.1.5/cat9k_iosxe.bin flash:
Switch# verify /md5 flash:cat9k_iosxe.bin
# Execution:
Switch# request platform software package install switch all file flash:cat9k_iosxe.bin auto-copy
Switch# install add file flash:cat9k_iosxe.bin activate commit
# Verify:
Switch# show version
Switch# show boot
Before vs After showing Cisco Catalyst Switches CLI vs Catalyst Center GUI for "upgrading firmware" task
RESEARCH QUESTION

"How can technology empower domain experts to orchestrate complex infrastructure?"

Spanning 8 years of empirical research across MathWorks and Cisco, this research theme focused on investigating how technical specialists (network / security engineers, operators, and architects or embedded software engineers) transition from low-level scripting to high-level system orchestrations. By studying workflows across MATLAB, Simulink, Catalyst Center, and Meraki, this theme established interface paradigms that reduce cognitive friction, optimize system visibility, and increase efficiency

Stream 01

Mental Models

How do experts go about with their work today?

Stream 02

Opportunities

Identify use cases, pain points, opportunities for automation

Stream 03

Concept Testing

Evaluating early and late designs with end users to catch issues

Stream 04

Benchmarking

Track adoption, benchmark experience and identify friction points

03 · EMPIRICAL DEVELOPMENT

Research Program

Connected studies conducted progressively answered different aspects of the central research question.

Research Stream 01

Campus Automation

2021 - 2026
Question

"How to enable network engineers / architects configure and deploy their brownfield switches on Catalyst Center?"

Methods
InterviewsConcept TestingSurveysCard Sorting
Outcome

Prioritized set of features for MVP of Campus Automation on Catalyst Center

DiscoverDefineDesignDeliverQ4FY24/Q1FY25: WLAN Survey& Card SortingQ1-Q3 FY25: Site to NetworkCampus Auto EvaluationQ2FY22: SwitchingJourney MappingQ4FY22: SwitchingInterviews & Early TestingQ3FY23: Network ZoneSurvey (Int/Ext)Q1-Q2FY23: SwitchingConcept TestingQ4FY24+: Switching/WirelessEFT Single Device Mgmt
Research Stream 02

Motor Control Blockset (0 -> 1)

2017-2020
Question

"How to enable control systems engineers develop and deploy motor control algorithms on microcontrollers, FPGAs or SoCs?"

Methods
InterviewsSurveyDesign Sprint WorkshopsConcept Testing
Outcome

Prioritized set of features for MVP of Motor Control Blockset

Release & Research Cycle19aCARDInterviews19bInterviewsDesign SprintMCT Survey20aMCT LibraryDesign Refine20bMCB BetaUsability / PURE
04 · REPRESENTATIVE CASE STUDY

Top-down Tree Architecture vs Layered Tier Architecture

3 Months12 Enterprise Networking ProfessionalsInterviews

Background

Current top-down tree architecture in Catalyst Center is not scalable and does not provide sufficient visibility into deployment. Large percentage of brownfield users defaulting to CLI leading to significant loss of revenue

Research Questions

  1. How are customers organizing their networks today?
  2. What are some challenges with the current way of organization?
  3. Understand if a layered tier architecture alleviates some of the pains?
  4. Evaluate all migration paths from old to new architecture

Research Approach

InterviewsConcept Walkthrough
Global
Country
State
City 1
Building 1
Floor 1
Floor 2
Floor 3
Building 2
Floor 1
Floor 2
Building 2
City 2
Building 1
Floor 1
representative illustration of top-down tree architecture in Catalyst Center

Key Findings

Due to NDA and intellectual property considerations, specific qualitative findings and metrics are generalized or omitted.

Research Contribution

This research validated requirements, settled a long-standing architectural debate, and evaluated the migration paths for coverage and design.

05 · PROGRAM SYNTHESIS

What We Learned

Enduring common principles derived from this research theme:

Match with the real world

Domain experts would like the automated processes mimic closely with how they do in the real world

Visibility into what automation pushes

Both network/security and embedded world experts would like to first see, validate what code is pushed into the hardware before actually deploying

06 · RESEARCH ASSETS

Research Artifacts

Empirical assets and frameworks generated to guide future enterprise-wide design and engineering direction.

Journey Maps

A comprehensive step by step flow for complex workflows - catalyst center, motor control etc.

Benchmarking Instruments

SUS scaling questionnaire to measure usability improvements

Card Sorting

Affinitized data points for designing effective information architectures

Prioritization Matrices

Kano survey on list of features generating prioritization to help with roadmaps

Wish Lists

A list of new features and enhancements requested by customers

Interview Framework

Semi-structured protocol used to query expert mental models of automation.

07 · PROGRAM PORTFOLIO

Additional Studies in this Research Program

This representative deep dive is one part of a much larger program of research.

Study 01

Unified Networking Personas

Methods: Workshops, Interviews, Secondary Research, Surveys
Outcome: A verified list of unified networking personas used across all of Cisco's enterprise networking
08 · PROGRAM OUTCOMES

Research Impact

The structural, organizational, and methodological contributions generated by this program.

Product Contributions

  • Validated customer journeys
  • Prioritized enterprise roadmap
  • Identified high-value automation opportunities

Organizational Impact

  • Artifacts shared across multiple product teams
  • Created common vocabulary (personas)
  • Aligned design decisions

Methodological Value

  • Introduced new metrics for journey maps
  • Introduced new prioritization methods like Kano survey
  • Pioneered usability benchmarking with SUS