AI integration · From $799 · Remote for US, UK and Gulf product teams

Add AI to Your Existing App — One Production Feature in One to Two Weeks

Your product works and customers pay for it. Now the board, the sales team or the customers themselves are asking where the AI is. You do not need a rebuild, a new platform or an AI team. You need one feature that earns its place, built into the codebase you already have, without stalling the roadmap your own developers are busy with. I'm Md. Emamul Mursalin, a senior AI and full-stack developer, and adding AI to existing products is a fixed-price package I offer to teams in the US, UK, Saudi Arabia and the Gulf.

The reason to start with one feature is that it is the cheapest way to learn the three things nobody can tell you in advance: whether your users actually use it, what it costs to run per user, and how good the answers are on your real data. Once those numbers exist, the decision about a second feature, or a full retrieval system, makes itself.

Md. Emamul Mursalin · Rajshahi, Bangladesh · remote worldwide · UTC+6 · Updated

What it costs to add AI to an existing app

The Starter package is $799 and takes one to two weeks. It covers one AI feature, either a chatbot, smart search, summarisation or content generation; prompt design with test cases; integration into your existing codebase; API key management with cost caps; basic usage logging; and documentation with a handoff call. It does not include retrieval over private documents, workflow automation, a UI redesign or model fine-tuning, and those exclusions are written into the scope so the price holds.

When the feature has to answer from your own data or automate real operational work, the Professional package is $2,499 over three to five weeks: retrieval over documents, a help centre or a database, up to three automated workflows with human review steps, up to two integrations such as Slack, HubSpot, Zendesk, Notion or Google Workspace, an evaluation set, a cost dashboard, guardrails with PII handling and two weeks of support. Enterprise is $6,999 over six to ten weeks for an internal AI gateway with budgets and an audit log, private model access through AWS Bedrock or Azure OpenAI, and assistants for up to two teams.

For comparison, a senior AI developer in the US or UK bills $150 to $250 an hour, so $799 buys less than one day of their time. If you would rather buy hours against an existing backlog, I bill $60 to $90 an hour.

How the work fits into a codebase that is not mine

The first two days are reading, not writing: your repository, your deployment process, your conventions and the place in the product where the feature will live. I work in a branch in your repository and open pull requests your team can review, so nothing reaches production that your own developers have not seen. My home stack is TypeScript with Next.js, React, NestJS and Node.js on PostgreSQL, and I have shipped on Firebase and MongoDB as well.

If your product runs on something else, such as Rails, Django or Laravel, the AI layer does not have to live inside it. It can run as a small separate service behind an API that your app calls, which keeps model keys off the client, puts cost caps and logging in one place, and lets your team keep working in the language they know.

  • Work happens in your repository, through pull requests your team reviews
  • Model API keys stay on the server, never in the browser or the mobile app
  • A hard monthly cost cap and per-user limits from the first deploy
  • Prompt test cases that run again whenever the prompt or model changes
  • Usage logging so you can see who uses the feature and what it costs
  • Documentation and a handoff call so your developers own it afterwards

Which AI feature is worth building first

The best first feature removes a step your users already find tedious. Summarising a long record before someone opens it. Drafting the reply, the description or the report that a user would otherwise write from a blank page. Searching by meaning rather than exact keywords. Answering how-to questions inside the product. Each of these is one feature, fits the Starter scope, and can be measured by whether people keep using it.

The features to avoid first are the ones that act without review on anything that cannot be undone, such as sending emails to customers or changing billing. Those belong in a later phase with human approval steps, which is what the Professional package adds. Part of the free first call is me telling you which of your ideas is a one-week feature and which is a three-month project wearing a disguise.

AI I have added to real products

CakeOrFake is the closest example to this kind of work. It is a Chrome extension that places AI directly inside an interface I do not control, Upwork's recruiter dashboard. It reads freelancer proposals, assigns risk scores, flags warning signs and writes the scores back into the page, with 35 cloud functions behind it, sign-in through Google, Facebook and Apple, and subscription billing through RevenueCat and Stripe.

Contently AI puts OpenAI, Anthropic and Groq behind one interface so a marketing team can generate and schedule social posts, which is the same model-routing pattern that keeps an added feature affordable. Virtual Client uses Gemini Live for real-time voice and OpenAI and Anthropic models for rubric-scored feedback, in production across several organisations with roughly 2,500 practice sessions delivered.

Working across time zones: US, UK and Saudi Arabia

I work remotely from Bangladesh at UTC+6, with no daylight saving. For a UK team that means I am online through your whole afternoon. For the US East Coast I overlap your morning until around noon and again in your evening, and we fix one weekly call that never moves. For Saudi Arabia the fit is unusually good: Riyadh is three hours behind me and we share a Sunday to Thursday working week, so your Sunday is a working day for me too.

The paperwork is what your finance and legal teams expect: a short contractor agreement with scope, price, confidentiality and full IP assignment, or your own NDA and contractor template if you have one. If your company must hire through an employer-of-record or contractor platform such as Deel or Remote, that works. Payment is by milestone in USD through Wise, Payoneer, bank transfer or a Stripe invoice.

Stack I work with

  • OpenAI API
  • Anthropic Claude API
  • Groq
  • Vercel AI SDK
  • Next.js
  • React
  • NestJS
  • Node.js
  • TypeScript
  • PostgreSQL
  • Firebase

Shipped work like this

Projects built with the same stack and patterns.

Ayana · HR Tech / Recruitment Automation

CakeOrFake — AI-Powered Upwork Applicant Screening Chrome Extension

AI-powered Chrome Extension that turns hours of Upwork applicant screening into minutes

Cloud Functions
35 endpoints
AI Model
GPT-4
Extension Size
1.5 MB
Read the case study

Personal product · Marketing / Content SaaS

Contently AI — Multi-Model Social Content SaaS

AI‑powered social content built fast, published faster

Pages built
42
API endpoints shipped
12
Monthly active users
3,200
Read the case study

Ayana Dev Studio · EdTech / Corporate Training

Virtual Client — Multi-Tenant AI Voice Coaching Platform

Teams rehearse live conversations against AI personas and get rubric-scored feedback.

Practice sessions
~2,500
Rules unit tests
90+
Collections hardened
8
Read the case study

What clients say

This project was handled in an excellent way. The seller understood the task clearly. Progress updates were helpful. The final output was accurate and polished. Everything arrived as promised. Glad to recommend this seller.

Alexi Denise

Individual client

He successfully completed all my tasks quickly and with excellent quality.

Imran Mohammed

Great work. Great communication, and a quick turnaround. I'd definitely order again.

Agrossen

Packages that fit

Fixed price, fixed scope. Starter and Enterprise tiers are on the pricing page.

Professional

AI Integration & Automation

$2,499/project

Launch offer$1,99920% off for the first 5 clients

3-5 weeks

RAG over your data plus automated workflows

Professional

AI Strategy & Fractional CTO

$1,499/project

Launch offer$1,19920% off for the first 5 clients

2 weeks

AI readiness audit and implementation roadmap

Professional

Full-Stack Web & SaaS Platforms

$2,499/project

Launch offer$1,99920% off for the first 5 clients

4-6 weeks

Production web app with backend and billing

Frequently asked questions

How much does it cost to add AI to an existing app?

With me, $799 for one AI feature such as a chatbot, smart search, summarisation or generation, integrated into your codebase in one to two weeks. Answering from your own private data plus workflow automation is $2,499 over three to five weeks. Model API fees are separate and estimated in the proposal, with a hard cost cap built in.

Do we need to rebuild our app to add AI?

No. An AI feature is an API call, a prompt, some guardrails and a piece of interface. It is added to the product you have, in a branch of your repository, through pull requests your developers review. If your stack is one I do not work in daily, the AI layer can run as a small separate service your app calls.

How long does it take to add an AI feature?

One to two weeks for a single feature. The first two days go on reading your codebase and agreeing exactly where the feature lives; the rest is the build, prompt test cases, cost caps, logging and a documented handoff.

What will the AI feature cost to run each month?

It depends on usage and the model, which is why every proposal includes a running-cost estimate based on your user numbers. I choose the cheapest model that passes the test cases, cache repeated requests, and set per-user limits and a hard monthly cap so the bill cannot run away.

Will you work with our in-house developers?

Yes, that is the normal arrangement. I work in your repository and your Slack, follow your code conventions and review process, and leave documentation and a handoff call so your team can maintain and extend the feature without me.

Is our customer data safe when it is sent to an AI model?

The business APIs from OpenAI and Anthropic do not train on your data by default, and I set the accounts up in your company's name so the data agreement is yours. Keys stay on the server, personal data can be stripped before a request is sent, and where policy forbids the public APIs the Enterprise package uses private model access inside your own cloud account instead.

What if we need more than one feature?

Start with one, measure it, then decide. A second feature can be another Starter, and once the work involves your own documents or automated workflows the Professional package at $2,499 is better value. For ongoing work there are monthly retainers from $500 or hourly work at $60 to $90.

Tell me what you're building

A 30-minute call is enough to tell you which package fits, what I would cut, and what it will cost to run. No pitch, no obligation.

Or email hello@mursalinsdesk.com