Hoshang Sheth logoHoshang Sheth
All Services
Service Deep Dive

Gen AI Applications.

AI that creates, from words to code to visuals - built around your business.

Book a Discovery Call Ask a question
At a glance
  • Timeline · 6-10 weeks
  • Investment · ₹45,000 - ₹80,000
  • Model · Fixed scope, fixed price
  • Ownership · Full source, no lock-in
Day 1
drafting content in your brand voice from day one
<30s
typical time to generate a first draft
₹1.5-8K
typical monthly AI running cost for an SMB workload

Figures from founder-built products and published benchmarks. Client results vary by scope, data quality, and use case.

The Problem

Why this matters right now.

Your team spends hours drafting content by hand, writing repetitive code, or paying for stock visuals one at a time. Generic AI tools produce content that sounds off-brand, code that ignores your conventions, and visuals that need reworking. What you need is generation built around YOUR templates, YOUR codebase, and YOUR voice.

The Approach

How I build this.

  1. 1We start by mapping exactly which content, code, or visual assets your team produces on repeat, and where the source data or examples for each one live.
  2. 2Your existing templates, codebase conventions, or brand guidelines are turned into a style guide the AI follows, so output looks and sounds like you made it, not like generic AI.
  3. 3The AI drafts from your structured data, past examples, and conventions, rather than inventing details, so facts, formatting, and style come from your actual business.
  4. 4Every draft lands in a review queue before it goes out, so nothing reaches a client, gets published, or gets merged unchecked, at least until you're confident enough to relax that step.
  5. 5The system goes live with monitoring in place, and you receive full documentation and a walkthrough before handover.
What you get

Deliverables

  • Marketing and campaign copy generated at scale, on-brand every time (product descriptions, ad variations, social captions)
  • Sales proposals, quotes, and client-ready documents drafted automatically from your pricing and templates
  • A conversational chatbot for your website or app that engages visitors in your brand voice, no document knowledge base required
  • A coding assistant trained on your codebase conventions, speeding up repetitive development work for your team
  • Image and video generation for product visuals, ad creative, and marketing assets
Under the hood

Technology stack

GPT-4o / Gemini 3.6 FlashClaude Code / CodexDALL-E 3 / RunwayLangChainFastAPINext.js

Model chosen per task - GPT-4o or Gemini for nuanced brand-voice writing, Claude Code or Codex for coding assistants, DALL-E 3 or Runway for image and video generation. Structured output schemas keep every generated document consistently formatted.

Case Study

SchemaHealer - AI-powered CRM data recovery

Problem

CRM datasets often arrive with inconsistent, renamed, or missing column names. Manual schema mapping is time-consuming, error-prone, and can result in incorrectly mapped data being passed into downstream systems.

Solution

Built an AI-powered schema recovery pipeline that combines deterministic matching, fuzzy matching, and Gemini 3.6 Flash semantic recovery to map inconsistent source columns to a canonical CRM schema. The recovered DataFrame is then verified against the expected schema, with confidence, healing, and verification results surfaced before generating the final import-ready CSV.

FastAPIGemini 3.6 FlashRapidFuzzPydanticNext.js
· MVPLive GitHub
Investment

₹45,000 - ₹80,000

Fixed scope, fixed price

Priced against the manual work it replaces, not just the build effort - a system drafting a hundred product descriptions or proposals a month is worth more than its build cost.

Every project is quoted as a single fixed price once scoped. The range above reflects how much project size varies - not variable billing within a project.

Typical Delivery
6-10 weeks
Payment Terms
50% to begin · 20% at prototype · 30% before handover
Frequently Asked

Questions specific to gen ai applications.

Whichever fits the task, budget, and compliance profile. GPT-4o or Gemini 3.6 Flash for brand-voice-sensitive writing, Claude Code or Codex for coding assistants, DALL-E 3 or Runway for image and video. I benchmark on your actual content before committing to one.
Anything factual, pricing, specs, dates, quantities, is pulled from the structured data you provide rather than left to the model to invent. Every draft is flagged for human review before it goes out, so nothing reaches a client unchecked until you're confident enough to relax that step.
No, and this matters for pricing. The chatbot here is a conversational interface that talks in your brand voice, drafts responses, and can guide a conversation, it doesn't look up your documents or company knowledge. If you need it to answer specific questions from your SOPs, policies, or product docs with source citations, that's the RAG & AI Agents service instead. Tell me on the discovery call what you need it to know, and I'll point you to the right one.
You own the code, the generated assets, and any prompts. LLM inference goes through your own API keys, you can rotate providers at any time.
For most SMB workloads I ship, ongoing cost lands between ₹1,500-₹8,000/month depending on volume. I always model the unit economics before we build.
No. Every system is built on your own API keys, your data goes to the LLM provider for inference only, not for training. If you need fully air-gapped deployment with no data leaving your infrastructure, that's a scope conversation we can have on the discovery call.
Next Step

Ready to scope gen ai applications for your business?

Not sure if your use case fits? Describe the repetitive work on the call and I'll tell you whether AI is the right solution and what it would actually take to build.

Book a Discovery Call See all services