KS
Founder & Engineering
Built forecasting systems for manufacturing planning teams. Writes prompt/ed on the AI hype hangover and where the actual value sits.
About Tensor AnalyticsA planning intelligence company
Webuildthe
opsplatformmost
ERPsdon'thave.
Tensor Analytics was founded to close the gap between spreadsheets and ERPs, not by adding another dashboard, but by building a connected operational data platform with ML forecasting, S&OP, automation, and AI agents that run the whole cycle through chat. Every action is human-approved. Every action is audited.
Our mission
Why Tensor exists.
Demand is planned in spreadsheets. Sales and supply chain argue about whose forecast to use. Supply plans don't reflect the latest demand. Exceptions surface too late. The plan keeps changing after production has started. There's no audit trail for who changed what, and no single view of forecast, supply, and gaps.
ERPs don't fix this, they weren't built for planning. BI tools show what happened, not what to do next. Enterprise IBP suites cost millions and take years to deploy. The result is gut-feel buffers, rigid policy rules, and S&OP cycles that take days because each team is reconciling a different version of the truth.
Tensor Analytics was founded to close that gap, not by adding another dashboard, but by building GrepEye: a connected operational data platform with eight capability pillars (data, rules, analytics, IBP, AI agents, forms, access control, config packs). ML ensembles run the math. The S&OP console enforces the workflow. AI agents handle the routine work through chat. Every action is human-approved and audited.
Operating principles
Six principles that shape every decision we make.
These aren't marketing claims. They're the operating constraints we use when we build, sell, and support GrepEye.
01
One operational data model, eight pillars
GrepEye isn't a forecasting tool bolted onto a dashboard. It's one platform, datasets, rules, analytics, IBP, AI agents, forms, access control, and config packs, all sharing a single operational data model. No stitching. No sync. One source of truth.
02
Planners spend time on judgment, not clicks
The ML ensemble runs the math. The S&OP console enforces the workflow. AI agents handle the routine triage through chat. Planners review exceptions, apply overrides with rationale, and approve agent actions. The judgment is theirs. The busywork is automated.
03
The agent proposes, the planner disposes
AI agents can read any dataset, view, projection, or dashboard. They can propose forecasts, supply netting, exceptions, overrides, and board packs. But every write action needs human approval before it lands. No silent automation. Every action is audited.
04
Sequential gates, not parallel chaos
The S&OP console enforces order: Demand Review unlocks Supply Review, which unlocks Exec S&OP. Lock the cycle once all gates clear, the plan freezes permanently. No more post-sign-off edits. No more finance and operations working from different numbers.
05
Indicative metrics, not vanity numbers
We don't promise 95% forecast accuracy on day one. We give you MAPE, WMAPE, and bias per SKU per model run. The ensemble leaderboard shows which model wins per product family. You know exactly where the model is reliable and where to apply human judgment.
06
Audit trail on every action
Every create, update, delete, and approve, by human or agent, is logged with actor, timestamp, and resource ID. The append-only ledger captures every operationally significant event with idempotency keys. Compliance isn't a feature toggle; it's structural.
Who's behind Tensor
A small team of planners, engineers, and writers.
Tensor Analytics is built by people who've run planning teams, written forecasting models in production, and sat in S&OP meetings where the loudest voice won. We're not ex-McKinsey. We're operators.
AP
Research & Editorial
Covers AI regulation, ethical AI in industry, and the gap between conference keynotes and factory floor reality. Leads prompt/ed editorial.
TA
Product & Planning
Former demand planner turned product. Owns the IBP pipeline design and the exception-first workflow that drives GrepEye's UX.
We're hiring
Building planning tools for manufacturers?
We're always looking for planners, engineers, and writers who've lived the spreadsheet-to-ERP pain. If that's you, write to us at teams@tensoranalytics.ai.
Get in touch
Let's build the ops platform your ERP doesn't have.
Book a 45-minute walkthrough. We'll run the full GrepEye platform on a representative slice of your data, connectors, ML ensemble, S&OP console, scenario planning, and AI agents in chat. No slideware.