Disclaimer: Independent product concept by Kaushal Khodifad. Not a live commercial product.return to portfolio
Disclaimer: Independent product concept by Kaushal Khodifad.

Roadmap & go-to-market

Land on safety. Expand to operations. Own the question.

Each release is gated by a learning question, not a date. Safety unlocks the first contract; operations ROI drives the renewal and the expansion.

Release plan

MVP → v0.5 → v1.0

MVPTrust the interface

Ask what happened

Prove a non-technical EHS user will ask questions and trust grounded answers enough to change behavior. Single site, retrospective.

RTSP ingestion on one edge node
Core safety events: PPE, LOTO, restricted zone, forklift proximity
NL query + grounded answers + clip citation
Anonymization, RBAC, audit log
Evals harness v1
Gate: At 2+ design-partner sites: EHS users reach 5+ grounded questions/week, faithfulness >= 0.95, and at least one documented 'we caught something because we could ask.'
v0.5Prove the ROI

Prove it pays

Demonstrate hard-dollar operational ROI in the same deployment and shift from reactive to proactive.

Operational ontology: idle, micro-stoppage, cycle complete
Weekly safety + ops digests
Proactive low-false-positive alerts
Dollarized loss ranking
First MES/EHS read-integration
Gate: 3+ quantified operational loss sources per site with credible $ impact, and one partner verbally commits to expand from safety to operations. North Star >= 20.
v1.0Survive procurement

Platform of record

Deployable across an enterprise's plant network; clears enterprise security + EU-AI-Act/works-council review by default.

Multi-site rollup + cross-plant query
Exportable DPIA / AI-Act conformity pack
Hardened RBAC, SSO/SAML, residency + retention
Evals v2 with online monitoring
SLAs + partner-installable edge appliance
Gate: One multi-plant enterprise live across 3+ sites; 100% security-review pass with no raw-video egress; NRR signal > 120%. North Star >= 30.

Now / Next / Later

Now · build the wedge

Edge ingestion + core safety extraction

NL query with grounded, cited answers

Privacy core + evals harness v1

2-3 design partners on retrospective safety

Next · prove the platform

Operational ontology + dollarized loss ranking

Proactive alerts + weekly digests

First MES/EHS read-integration

Multi-site rollup; SSO; conformity pack

Later · own the category

NL custom-event definition

Agentic actions + near-real-time severe-hazard alerts

Deeper closed-loop ops; air-gapped option; industry event-pack marketplace

The two-sided value engine

One sensor estate, one event layer, two budget owners

Safety (budget-unlocker)

BuyerEHS / Safety Director
Existing budgetYes: EHS + insurance + compliance
DriverLiability, regulation, 'never again'
Sells theFirst contract

Operations (ROI-justifier)

BuyerPlant Manager / Ops / CI lead
Existing budgetHarder; must be proven
DriverThroughput, OEE, cost-per-unit
Sells theRenewal + expansion

Go-to-market

Who we sell to

Mid-market to enterprise discrete + process manufacturers (automotive, electronics, food & beverage, pharma, logistics) with multiple cameras already installed, an EHS function, and an active safety/compliance mandate. Multi-site networks for expansion. Often a recent incident or an insurance/regulatory trigger.

The buying committee

EHS / Safety Director
Cares about: Incidents, liability, proving ROI
Ask what happened; prove what's working.
Plant Manager / Ops
Cares about: OEE, throughput, downtime cost
Ask your cameras why the line slowed.
IT / Security
Cares about: Data egress, integration, RBAC
Raw video never leaves the site.
Legal / Compliance
Cares about: GDPR, works council, EU AI Act
Anonymized, auditable, built for the Act.
CFO / Plant GM
Cares about: Payback, risk-adjusted ROI
Safety savings + recovered output, one platform.

Pricing & sequencing

Pricing logic

Per-camera + per-site platform fee, tiered Safety -> Safety+Ops -> Enterprise/compliance. Per-camera scales with estate size; the platform fee captures the query layer.

Anchor on value, not compute: price against avoided incident cost plus recovered output. Incumbents report ~77% injury reduction and ~$1.1M annual savings at a single large site.

Pilot pricing kept low to remove the entry barrier; expansion pricing captures platform value once ROI is proven.

First 4 quarters of selling

Q1

2-3 design partners on safety retrospective. Obsess over the ask-and-find proof and faithfulness.

Q2

Convert to paid; turn on ops ontology; produce 2-3 ROI case studies.

Q3

Open the channel (1 SI + 1 EHS platform); start EU 'built for the Act' outbound.

Q4

Land first multi-plant enterprise; publish the 'query layer for industrial video' POV.

Risks & mitigations

Bought VLM under-performs on factory events
High

Validate per-event in the first design partner; build a fine-tuning path; keep the model layer portable.

Safety incumbents add an NL query feature
High

Move fast on interaction + ontology depth and trust (evals, conformity), which are hard to copy quickly.

Works-council / EU AI Act blocks deployment
High

Privacy-by-design + conformity pack as P0/P1; 'built for the Act' as a wedge, not a tax.

Hallucinated safety answers destroy trust
Critical

Grounded-or-silent principle; faithfulness gate in evals; clip citation mandatory.

Long enterprise sales cycle starves revenue
Medium

Low-friction paid pilots; channel leverage; insurer co-sell.

Foundation-model supplier dependency/price
Medium

Abstraction layer; multi-model; option to bring inference in-house at scale.

↑ Independent product concept by Kaushal Khodifad. FactoryLens is an independent product concept by Kaushal Khodifad; it is not a real company or a commercial product. It explores the natural-language intelligence over factory video space. Not a live commercial product. Data is illustrative.

Return to portfolioOpen the live demo