Innovation Management · Foresight Research · Saarbrücken, DE

I research where industries are heading - and find the AI use cases worth building.

Innovation management and foresight researcher at the Applied Innovation Lab of the August-Wilhelm-Scheer-Institut. I go deep on a business first: interviews, workshops, scenario methods. Then I identify and prototype the AI use cases that fit how it runs. Increasingly, I build them myself.

M.Sc. Global Foresight & Technology Mgmt M.Eng. Wirtschaftsingenieurwesen (in progress) Agentic AI workflows in daily use EN · DE

Selected work · 2024-2026

Case files

A mix of client engagements and personal systems. Client work is anonymized; the personal projects run on my machines every day.

CASE 01 · PERSONAL

PythonWhisper STTClaude agentsElevenLabs

Running daily

FRIDAY, a voice assistant that actually does the work

A JARVIS-style voice AI, built end-to-end and always on.

Problem

I wanted hands-free, conversational access to my entire project knowledge base. Not another chatbot in a browser tab.

Built

Wake-word activation, local speech-to-text, Claude reasoning with my knowledge vault as memory, and natural voice replies. On top of that sits a work mode: one spoken instruction kicks off a real autonomous coding task.

Outcome

Runs always-on on my Mac. I hold full spoken conversations about my projects and hand off real tasks by voice.

CASE 02 · PERSONAL

ObsidianClaude CodeAgent workflowsDashboards

Running daily

Cortex, an AI-operated personal operating system

Everything I work on, tracked and partly run by agents.

Problem

A dozen parallel workstreams (research, client projects, university, applications) and no single place where their state lived.

Built

A knowledge vault operated by AI agents: living mission-control dashboards with project lanes and timelines, automated session notes, and a morning triage agent that reads my inbox and pre-sorts opportunities before I wake up.

Outcome

One glance shows the state of everything. The boring parts of staying organized no longer cost me time.

CASE 03 · CLIENT · ANONYMIZED

Market researchCompetitor analysisBusiness model

Letter of intent signed

AI regulation monitoring for a German specialist publisher

The market research behind a new AI product.

Problem

Compliance teams at mid-sized companies drown in regulatory change. A specialist publisher wanted to know whether an AI product could carry that load, and whether anyone would pay for it.

My part

I ran the market research: market sizing, competitor analysis and business-model options for the product. The findings fed the go/no-go decision and the product's positioning. The prototype itself was built by the engineering team.

Outcome

Prototype validated with target users; an established publisher signed a letter of intent.

CASE 04 · CLIENT · ANONYMIZED

Agent orchestrationHuman-in-the-loopCRM

In active use

Agentic lead research with human-in-the-loop gates

Multi-agent pipelines that find real people, and a human who approves every record.

Problem

B2B teams need named, verified decision-makers at target companies. Manual research doesn't scale; fully automated research can't be trusted.

Built

A multi-agent research pipeline with strict pass/fail validation criteria (current employer verified, role matched by what the person does rather than by title keywords) and a mandatory human review step before anything is written to the CRM.

Outcome

Verified contact discovery at a fraction of the manual effort, without sacrificing data quality. The human-in-the-loop design is the point, not a compromise.

CASE 05 · UNIVERSITY × INDUSTRY

IoTMQTTLive sensor data

Presented at Pixida, Munich

Eco-Score dashboard on live IoT sensor data

Making the energy behavior of real devices visible in real time.

Problem

Device energy consumption is invisible until the bill arrives. The industry partner wanted a live eco-score that a non-engineer could read, instead of raw telemetry.

Built

A real-time dashboard that scores device energy behavior from live sensor streams over MQTT. Team project, delivered against a running backend.

Outcome

Presented to the industry partner at Pixida in Munich, 2026.

How I work

Understand deeply, then ship

The business comes first

Stakeholder interviews, on-site workshops, and the foresight toolkit: Delphi studies, scenario analysis, horizon scanning. Before any system gets built, I can explain the business to the people who run it.

II.

Prototypes over slideware

Working AI workflows in weeks instead of strategy decks in quarters. A demo the client can touch settles arguments that meetings never will.

III.

Humans stay in the loop

Automation with validation gates: agents do the volume, people make the calls that matter. Systems earn trust by being checkable, not by being impressive.

Background

Engineer by training, futurist by degree, builder by habit

I started as an automobile engineer in India, moved to Germany for a master's in Innovation Management with a focus on Global Foresight & Technology Management, and stayed. Since then I've worked on the production floor of a German carmaker and run Delphi studies at a foresight institute. My master's thesis ended up feeding directly into an insurer's digital strategy.

Today I work in the Applied Innovation Lab at the August-Wilhelm-Scheer-Institut in Saarbrücken, running innovation and AI-strategy engagements with industry clients. Alongside the job I'm completing a second master's in Wirtschaftsingenieurwesen (industrial engineering and management). Outside work hours I build the systems I wish existed, from voice assistants to agent-operated dashboards. The two halves feed each other: research tells me what's worth building; building keeps the research honest.

Education: M.Eng. Wirtschaftsingenieurwesen, TH Ingolstadt (in progress)

M.Sc. Global Foresight & Technology Mgmt, TH Ingolstadt

B.Tech Automobile Engineering
Certification: CAPM, Project Management Institute
Languages: English, German

Hindi, Gujarati
Based in: Saarbrücken, Germany
Next build: Agentic automation for the innovation-management process

Contact

If you're building something that needs both the map and the machine, talk to me.

Write me about AI transformation, agentic systems, or a problem that doesn't fit a category yet. The classic CV lives here too.

Email me LinkedIn Download CV (PDF)