AI assistants & agents
AI agents that do real work, with a person on the approval.
- from ₹50,000two-week sprint, ends in a fixed quote
We design and build AI agents that read your messages, documents and systems, decide what to do, and act through your tools: CRM, WhatsApp, email, spreadsheets and internal APIs. Anything a customer will see waits for a person on your team to approve it.
Right for you if
Who this is for.
- Teams with a repetitive, rules-heavy process that eats hours every week
- Founders who want an agent at the core of a new product
- Companies that tried a chatbot and need something that actually completes tasks
Every agent project starts with a two-week sprint that ends in a fixed quote.
What you get
What an agent project delivers.
Process map and agent design
Which steps the agent owns, which need a person, and what it may never do.
Tool and data integrations
Connections to your CRM, WhatsApp, email, databases and APIs, including MCP servers where useful.
Approval queue
Drafts wait for a human before anything reaches a customer. Approvals and edits are logged.
Evaluations
A test set of real cases the agent is scored against before launch and after every change.
Model routing
Claude, OpenAI or open models per task, with fallbacks if a provider is down.
Run and improve
Monitoring, cost tracking per task, and a monthly improvement cycle on a retainer if you want it.
Costs
What an AI agent costs, and what runs after launch.
| Stage | What it covers | Starting price |
|---|---|---|
| Sprint, 2 weeks | Process map, prototype, approval rules, fixed quote | from ₹50,000 (credited) |
| A focused agent | One or two of your tools, approval queue, test set | from ₹5 lakh |
| Agents across several systems | More integrations, security review, support | quoted after the sprint |
| Running and improving | Monitoring, cost per task, monthly improvements | from ₹30,000 a month |
Model usage and hosting are separate and estimated in the sprint. Starting points, not quotes; prices exclude taxes.
Read the full cost guide →Examples
Agents that take work off a team.
Illustrations of what we build, not client claims.
- A sales agent that qualifies inbound WhatsApp leads, updates the CRM and drafts the reply for approval
- An operations agent that reads order emails, checks stock and prepares dispatch notes
- A research agent that compiles weekly competitor and market briefs from sources you choose
- A support agent that answers from your documentation and hands off to a person with full context
How it works
Every project starts with a two-week sprint.
Sprint · 2 weeks
Interviews, a clickable prototype, scope, architecture and a fixed quote. From ₹50,000, credited to the build.
Build
Weekly releases you can try. Fixed scope, milestone billing.
Launch
Production release, monitoring and launch fixes.
Run (optional)
Retainer or care plan to run and improve it.
Built with
- Claude
- OpenAI
- Open-source models
- MCP
- Node.js / TypeScript
- Python
- PostgreSQL
- Redis queues
Proof we run ourselves
WhatsApp Gateway
The messaging layer our agents use runs live on our own number.
See the product →Questions
AI assistants & agents: straight answers.
What is the difference between an AI agent and a chatbot?
A chatbot answers questions. An agent completes tasks: it reads context, decides on steps and uses tools such as your CRM or WhatsApp to act, with a person approving anything a customer will see.
How much does an AI agent cost to build?
Every agent starts with a two-week sprint from ₹50,000 that ends in a fixed quote. A focused agent on one or two of your tools usually starts from ₹5 lakh; agents across several systems are quoted after the sprint. Model usage and hosting are separate running costs.
Will the agent send messages to my customers on its own?
Not unless you decide it should. By default every customer-facing action waits for approval, and the approval rules are part of the design.
Do small businesses need AI agents?
Most do not start there. A small business usually gets more from automation or an AI receptionist first. An agent makes sense when someone copies information between tools every day, or leads and orders wait because nobody has time, and the job has clear steps.
Should we buy a ready-made AI tool or build an agent?
Buy when a tool already does the job your way. Build when the agent has to work inside your own systems and rules. The sprint ends with that recommendation, including when the answer is to buy.
Related
Other things we build.
Written scope, fixed quote
Tell us what slows your business down.
Websites and automation are quoted after a 30-minute call. App and software builds start with a two-week sprint that ends in a written scope and a fixed quote.
- 30-min call
- Written scope
- See the price list →