✦Field Notes · New Essays Monthly

Notes from the shipping floor.

What we've learned making AI ad creative at volume for D2C brands, and building the automations behind it: WhatsApp chatbots, AI voice agents and the dashboards that prove they work.

1,500+
Ads shipped
1,000+
Content pieces
100%
Source Handoff
A media buyer at her desk at night comparing a vertical ad on her phone with a grid of ad thumbnails on her laptop
★ FeaturedContent
Vaidehi Khunt
Vaidehi Khunt
CEO · Brand & Content
· 5 min read

Meta Andromeda changed what your ad account needs: not more ads, different ones

Meta's Andromeda retrieval engine treats near-identical ads as one. How D2C brands should plan creative diversity across persona, angle, format and setting.

Generative AdsContent At VolumeNo Shoot
Read essay→

All essays · 13

Sorted: newest first

A young Indian woman creator filming a UGC video in her bedroom with a phone on a tripod and a ring light, holding up a serum bottle
Content··4 min

What a UGC video really costs in India in 2026: creators vs AI UGC

Indian UGC creator rates run from about ₹2,000 to ₹50,000+ per video. The full cost once you add usage rights, shipping and reshoots, and how AI UGC compares.

Vaidehi KhuntVaidehi Khunt
→
Close-up of a woman's neck and collarbone wearing a delicate gold pendant necklace against a dark background
AI··4 min

AI product photography for jewellery and fashion: on-model shots without a photoshoot

How we turn a real ring, necklace or kurta into on-model photos without a shoot, and the checks that stop the AI from changing the product.

Tarun ChelumallaTarun Chelumalla
→
A woman's hand holding a smartphone with a WhatsApp-style chat open, on a sofa in the evening beside a glass of chai
Automations··5 min

WhatsApp chatbot for D2C brands: how we build one that qualifies leads 24/7

The five parts of a WhatsApp chatbot that answers and qualifies around the clock, the handoff rules that stop it annoying customers, and the dashboard that proves it works.

Tarun ChelumallaTarun Chelumalla
→
A man answering a phone call on an apartment balcony at dusk with city lights behind him
Automations··4 min

AI voice agent for Meta lead ads: call every lead within minutes and book the meeting

How an AI voice agent calls every Meta lead form submission within minutes, qualifies the lead in their own language, and books the meeting on your calendar.

Tarun ChelumallaTarun Chelumalla
→
A founder at a desk at night looking at a large monitor showing charts and dashboard tiles
Automations··3 min

AI automation dashboards: why every agent we ship comes with one

Workflow logs say 'success' while the business result is zero. What an automation dashboard should show: is it running, is it working, and is it worth it.

Tarun ChelumallaTarun Chelumalla
→
A developer typing at a desk at night in front of two monitors showing code
Automations··4 min

Claude Code automation: the guardrails that stop it breaking production

Scoped permissions, plan mode, hooks and the prompt-as-spec pattern we use to run Claude Code on client codebases without it rewriting production.

Tarun ChelumallaTarun Chelumalla
→
Top-down view of skincare bottles, a phone showing a vertical video and a row of printed photo thumbnails on a walnut table
Content··4 min

Our AI content engine for D2C brands: the 4-system stack, tool by tool

The voice fingerprint, brief engine, draft pipeline and distribution formatter behind our D2C content work, with the tools we use and the steps we never automate.

Vaidehi KhuntVaidehi Khunt
→
A man drawing a flow diagram of connected boxes and circles on a dark glass wall with a marker
Automations··5 min

n8n in production: the 7 nodes behind every workflow we ship

HTTP Request, Code, Edit Fields, IF, Loop Over Items, Merge and Error Trigger: the n8n nodes that hold production workflows together, and the mistakes each one prevents.

Tarun ChelumallaTarun Chelumalla
→
A hand pinning a photo onto a large board covered in dozens of photos of different people each holding the same serum bottle
Content··4 min

AI UGC ads at scale: how one brief becomes hundreds of ad variants

The brief-to-variant pipeline behind our bulk ad creative: persona, angle and setting as variables, the production stack, the review gates, and the cost in rupees.

Vaidehi KhuntVaidehi Khunt
→
Two people at a wooden table, one sliding an open laptop across to the other in a handover
Studio··3 min

Why we don't sell retainers (and what you get instead)

Retainers pay for time, not results. How we price automation as fixed-scope builds and ad creative as a fixed number of videos a month instead.

Vaidehi KhuntVaidehi Khunt
→
A hand pinning index cards into three neat columns on a dark cork board
AI··4 min

Prompt-as-spec: how to write AI agent prompts like engineering tickets

Input, constraints, acceptance criteria. The three-part prompt structure that makes AI agents behave the same way on the 500th run as on the first.

Tarun ChelumallaTarun Chelumalla
→
A smartphone lighting up with a notification on a bedside table at night beside a glass of water and glasses
Automations··4 min

Automation pipelines that don't break on the third launch

Idempotency keys, a retry policy that knows when to give up, human checkpoints and alerting. The unglamorous parts that keep n8n pipelines alive after launch.

Tarun ChelumallaTarun Chelumalla
→
An editor marking up printed pages with a red pen beside a magnifying glass on a desk
Content··5 min

AI slop: an 8-signal checklist for catching generic AI content before it ships

A short, blunt checklist for spotting generic AI content before it goes out under your brand, with the fix for each signal.

Vaidehi KhuntVaidehi Khunt
→

The Cinqa Dispatch

One short essay.
Every other Wednesday.

Pipelines, prompts, and what we shipped. No drip funnels. Unsubscribe in one click.

~1,200 founders + operators. Zero spam.

Or skip the reading

Ready to Grow?

Start a Project →Read the FAQ