Jitendra Singh · AI Product Manager

About

AI product manager with 8 years across product and program delivery. Launched two 0-to-1 products at Taggbox — UGC Suite and Tagshop.ai — and scaled the three products I managed from 30K to 65K+ users, $1.3M to $2.5M portfolio revenue with the team. Every AI feature I ship starts with ground truth, an eval set and an accuracy gate.

75 days concept to launch, Tagshop.ai
$1.3M → $2.5M portfolio revenue, as a team
30K → 65K+ users across three products
50+ paid enterprise customers, UGC Suite year one

Experience · 15 flights

The same work as the Departures board, in a familiar profile format.

TS075REALIZED

Tagshop.ai, idea to paying customers

Discover · 75 days · Launched

38+ paying

From the problem: Each ad took 2-3 weeks and a team of freelancers.
To the result: Ads in about 5 minutes at 20x lower cost. 1,000+ signups in week one, 38+ paying customers in 15 days.
My decision: I built the first working prototype myself to test the core flow, priced it against the category leaders, seeded it with existing customers and launched at D2C events. Concept to launch in 75 days.

Visit Tagshop.ai

UG120REALIZED

UGC Suite, wireframes to $120K

Onboard · 6 months · Launched

$120K yr 1

From the problem: Assets, website widgets and event screens lived in separate tools.
To the result: About $80K in the first six months, $120K in year one. 50+ paid enterprise customers in one year, at an average ticket size of $3,000.
My decision: Combined them into one platform with asset management, packaged by plan and sold to enterprise only. Built the onboarding flows and guides and tracked every step.

AI060MEASURED

AI moderation, renewals +10%

First value · 1-3 wks · Delivered

+10%

From the problem: Customers spent hours cleaning feeds instead of using them.
To the result: Renewals up 10%, churn down 5%, widget creation up 7-10% vs the previous month in the pilot, and conversion up 6% from plan gating. Manual moderation effort also fell 60%.
My decision: I owned the fix end to end with engineering: AI recommendations and duplicate detection, piloted plan by plan. Responses took 5-10s in testing but about 30s live under load, so we upgraded the GPUs before wider release.

DD020DIRECTIONAL

AI duplicate detection, one-click removal

First value · 2–3 months · Delivered

1 click

From the problem: Duplicate images in feeds, removed by hand.
To the result: Built alongside AI moderation for the same growth problem; its results are tracked together with AI060.
My decision: Detect duplicates automatically and offer one-click removal, with engineering.

PS095DIRECTIONAL

AI product search, Samsung LATAM

Grow · — · Delivered

No manual tagging

From the problem: Every image matched and tagged by hand.
To the result: Gemma, PaliGemma and vector search now identify, match and tag products automatically. It helped win many customers.
My decision: Automate the manual match of about 1,000 products per image.

AP018DIRECTIONAL

Autopilot, task automation

Habit · ~6 months · Shipped

3 products

From the problem: Repetitive manual actions every day.
To the result: Feed refresh, updating posts in widgets and removing duplicate images now run automatically, across all three products.
My decision: Built it first in UGC Suite, then copied it to the other two products.

BT020MEASURED

Brand auto-theming, 5 min to 20 sec

Onboard · 20 sec · Delivered

5 min → 20 sec

From the problem: 5 minutes of manual theme setup.
To the result: Setup from 5 min to 20 sec, fewer support requests and better feedback.
My decision: Enter a website URL and the product inherits the brand colours into the widget or social wall.

PY006MEASURED

Free to paid, +6% conversion

Pay · Pilot · On time

+6%

From the problem: Giving the feature away would have cost more than it earned.
To the result: Free-to-paid conversion rose 6%, and cost stayed tied to revenue.
My decision: We upgraded capacity and kept the feature to paid plans. Every later AI launch went through the same staged check, with experiments of 1-3 weeks on features and pricing.

GR250REALIZED

3 products to $2.5M revenue

Grow · 2.5 yrs · Arrived

$1.3M → $2.5M

From the problem: Three products competing for one team's time.
To the result: Revenue grew from $1.3M to $2.5M and users from 30K to 65K+, as a team.
My decision: I owned the roadmaps for all three, planned each quarter and retired low-performing features and integrations to cut cost. Led a team of up to 23 and mentored a junior PM.

ES003DIRECTIONAL

Enterprise AI search, live for 3 teams

Habit · 2026 · Live

3 teams

From the problem: Answers were spread across four systems.
To the result: One search across all of them, live for HR, finance and developer teams and pitched to customers.
My decision: Led the product: which systems to connect first, how answers cite their source, and who sees what.

Try my version: MyBuddy

KR080MEASURED

Warehouse cameras, live AI at docks

First value · 2025 · Shipped

~80% at launch

From the problem: Dock and warehouse activity was tracked by hand or not at all.
To the result: Live production camera projects at warehouse docks, about 80% accurate at launch.
My decision: Prioritised use cases by measurable efficiency gain rather than technical novelty, and wrote requirements from the floor problems up.

BR001REALIZED

Bravo, a free private AI agent

Build · Live · Live

$0 / month

From the problem: Good AI assistants cost $100+ a month or need a developer to set up.
To the result: A free, open-source desktop agent with a one-click installer, live today.
My decision: Runs on your own laptop by default, your own API key is optional, mail stays read-only, and it asks before anything risky.

Open Bravo

CL034MEASURED

Clearline, quotes you can trust

Build · Live · Live

34/34 checks

From the problem: Buyers compare quotes by hand and can't trust AI with the numbers.
To the result: One comparison where every figure traces to its source. 34/34 checks pass.
My decision: The AI reads and explains; the app does all the maths. Anything unclear goes to a person instead of being guessed.

Open Clearline

AK522DIRECTIONAL

Akshar Bench, AI tested by people

Build · Live · Live

522 votes

From the problem: No one knew which model to trust for non-English text.
To the result: 522 judgments from 8 raters: the best model scored 5/5, the others 2/5 and 0/5.
My decision: Before trusting any model, test it with real people on real use cases. Small sample, so I treat it as direction, not proof.

Open Akshar Bench

JS026Open seat

Your team, next

Your gate · Now · Boarding

Available now

From the problem: An open product role.
To the result: Remote or relocation. Available to start immediately.
My decision: Write with context and I'll reply with something useful. Email: jitendrashekhawat00@gmail.com · Phone: +91 85599 91626 · LinkedIn: linkedin.com/in/jitendra-singh-7b311b123

Message me on LinkedIn

Contact

Send a message — it goes straight to Jitendra's inbox.