AI WorkflowsPreview

A decision, delivered. Not just an answer.

A workflow chains a SQL pull, a Python transform, a chart and an AI step, then delivers the result to where your team works, on a schedule. It runs on your data, raw tables or transformed models alike, so you go as deep as the question needs.

Request early access See Ocular AI

A workflow, not a single answer

A SQL pull, a Python transform, a chart, an AI read, then delivery. Pick one and read it top to bottom, the way your team would build it.

Weekly margin & returns watch.ipynbRuns every Monday, 7:00 IST

01SQLPull CM2 and returns by state

select state, week, cm2, returns_value
from marts.contribution_margin
where week >= dateadd('week', -12, current_date)

state	cm2	Δ wow
Maharashtra	24.3%	-2.1
Karnataka	27.1%	+0.2
Delhi	25.8%	-0.4

02PythonFind the biggest weekly drop

piv  = cm.pivot('week', 'state', 'cm2')
worst = piv.diff().iloc[-1].idxmin()   # 'Maharashtra'
at_risk = returns_at_risk(worst)        # 320000

worst = 'Maharashtra' · at_risk = ₹3.2L / week

03ChartCM2 by week, Maharashtra

CM2 % · last 12 weeks

04AIExplain the drop

Returns in pincodes 400053, 400061, 400072 and 400080 on a single courier drove the fall. Switching courier recovers about ₹2.6L per week.

05DeliverSend the brief

Sent to#growthfinance@brand.com

Every cell works on your data, raw tables or transformed models, so you go as deep as the question needs.

Deliver where your team works

Slack

The decision lands in the channel, with the chart and the why.

Google Sheets

Numbers refreshed in the sheet your team already lives in.

Email

A scheduled brief in the inbox. No login needed to read it.

In-app

Or pinned to the Ocular team board, always current.

Automated distribution

Push reports to Slack, Sheets and email on the cadence the business runs on, so the insight reaches people where they already work.

Proactive AI alerts

It watches the model and pings you only when a number actually moves, with the driver already attached.

SQL, Python, no-code and AI

Mix a SQL pull, a Python transform and an AI step in one workflow. Go as deep as the question needs.

Connect any source

Pull from every channel you sell on, storefront to marketplaces to quick commerce, raw or cleaned and reconciled, ready for the analysis.

A notebook over all your data.

Bring SQL, Python, no-code blocks and AI into one workflow, over your raw tables and your transformed models alike. The data already carries the dimensions Indian commerce runs on, pincode, COD, RTO, festival seasonality and the rupee. AI can compose the whole workflow, or you build it block by block.

See the semantic layer

NotebooksOcular

SQL, Python, no-code and AI

Delivers to Slack, Sheets, email

Raw and transformed data together

India-native dimensions built in

AI composes the workflow

Built for commerce operators

What teams build with it

01Monthly business reviewP&L versus plan, channel mix, SKU movers and cohort deltas, assembled and delivered before the call.

02Margin leak investigationA CM drop decomposed to the pincode, courier or campaign that caused it, with the recovery quantified.

03Pre-festival replenishmentDemand forecast to cover days to order-by dates, ranked by the margin you would lose by stocking out.

04Channel reconciliationMarketplace-claimed numbers walked back to what landed in your P&L, every week, on its own.

Analysis that delivers itself.

AI Workflows is in early access. We scope your first workflows to the reviews you already run, then ship. Talk to us.

Request early access See Ocular AI