Susam Labs

AI systems that pay for themselves.

Susam Labs is an AI product studio. We design, build and ship AI-native software end to end — AI agents, voice agents, retrieval systems and agentic workflow automation on n8n — and we run hands-on AI training for engineering teams and institutions.

How working with us goes — entered as a business

Fixed scope · fixed price · you own the code

See fixed-price automations →
Explore programs →

Write, talk or call. Send a written brief, record a 30–90 second voice note, book a slot, or just message us on WhatsApp. A rough shape of an idea is enough — you do not need a spec to start the conversation.

Ways to reach us →
Book an intro call See the work →

Published prices. Fixed scope and a launch date before you commit. Full handover — you own the code.

What we build

  • Voice agent answers an inbound call and updates the CRM
  • Document Q&A grounded in a 1,200-page manual, with citations
  • Clinical note drafted from a consultation, clinician approves
  • n8n workflow enriches a lead, scores it and routes it to an owner
  • Yesterday's numbers land in WhatsApp as a written summary
  • Someone asks a question in WhatsApp and gets it answered from the data
  • Eval suite gates a prompt change before it reaches production
  • Agent plans a task, calls MCP tools, escalates to a human at the checkpoint

Every one of these is a system we have built and run in production.

Trusted by

10+
AI systems shipped and live

Every one of them listed on the work page

100%
of clients live and seeing a return

Every engagement we've taken is in production

100%
of code and workflows handed over and deployed

Your repo, your infrastructure, no lock-in

1 quarter
what we scope an automation to pay back in

Check it against your own numbers below

Start small, see the return first.

Productised automations: one named outcome, one fixed price, one delivery date. Most clients start with one and extend it once it has paid for itself.

Scope one →

n8n workflow

Kill the recurring task nobody wants to own

Someone spends four hours a week moving data between two tools.

Task automation on n8n for the recurring back-office work that never justifies a hire: invoice reconciliation, onboarding checklists, report assembly, data sync between systems that have no native integration. One workflow, one owner, one fixed price.

  • One process mapped, automated and documented end to end
  • Error handling and retries, with an alert when a run needs a human
  • Runs on your own n8n instance — you keep the workflows
  • A written runbook so your team can change it later

From $1,800 fixed· 1–2 weeks

OpenClaw agent

Get your business reports in WhatsApp

Nobody opens the dashboard, so nobody sees the numbers.

A WhatsApp reporting agent built on OpenClaw, the open-source self-hosted AI agent gateway. It pulls yesterday's numbers from your database, sheets or CRM on a schedule and posts a written summary to a WhatsApp thread your team already reads.

  • Scheduled daily, weekly or month-end summaries
  • Written in plain language, not a screenshot of a chart
  • Runs on your own infrastructure — data never leaves it
  • Ad-hoc queries: ask for a number and get it back in the thread

From $2,500 fixed· 2 weeks

n8n workflow

Every inbound lead enriched and scored before you reply

Your team researches each lead by hand, then replies a day late.

An n8n lead-enrichment pipeline. Every inbound form fill, download or reply is enriched against firmographic and contact data providers, scored against your ICP, routed to the right owner in the CRM, and announced in Slack — inside a minute, with no human step.

  • Enrichment from your existing data providers, deduped against the CRM
  • ICP scoring rules you can read and change without an engineer
  • Round-robin routing plus a Slack alert on anything above threshold
  • Full run history in n8n — every lead's path is auditable

From $2,900 fixed· 2 weeks

OpenClaw agent

Chat with your own data, in WhatsApp

Answers live in four systems and one person's head.

A conversational agent that answers questions about your own business data from inside WhatsApp. Built on OpenClaw with read-only connections to your systems, it resolves a question into a query, runs it, and answers with the figures and where they came from.

  • Read-only connectors to your database, sheets and CRM
  • Answers cite the source record, so numbers can be checked
  • Per-user permissions — people see only what they already could
  • Self-hosted, with a full audit log of every question asked

From $3,500 fixed· 2–3 weeks

n8n workflow

An outbound engine that fills the pipeline while you sleep

Prospecting stops the moment your team gets busy delivering.

An n8n lead-generation workflow that builds target lists from your ICP definition, verifies contact data, drafts a genuinely personalised first touch with an LLM grounded in the prospect's own site, and sequences follow-ups — with a human approving every send until you decide otherwise.

  • ICP-driven list building with deliverability and suppression checks
  • Per-prospect research pass, so the first line is about them
  • Human-in-the-loop approval queue before anything sends
  • Replies classified and pushed to the CRM as tasks or deals

From $4,200 fixed· 3 weeks

Custom orchestration

Whole workflows done end to end, not just chat answers

Your AI answers questions but still can't finish a job.

Multi-step agentic workflows: an orchestrator that plans a task, calls your systems through typed tools and MCP servers, checks its own work against evals, and escalates to a human at the points you choose. Built for processes with a definition of done, not for chat.

  • Tool and MCP server design against your real systems
  • Eval suite gating every prompt and model change before release
  • Explicit human-in-the-loop checkpoints and rollback paths
  • Tracing and cost observability from day one

Quoted after a scoping call

What would automation save you?

Put in the hours your team currently spends on the work you'd automate. The arithmetic is printed underneath, and nothing you type leaves your browser.

6 hours
2 people
₹400
Currency

You'd get back

₹1,74,720a year — about 437 hours of your team's time

Pays for itself in 10.9 months

against kill the recurring task nobody wants to own at ₹1,58,400, delivered in 1–2 weeks.

That's longer than the quarter we scope automations to pay back in. On these numbers we'd want a call before you commit — usually there's a bigger process worth automating first, and sometimes the honest answer is not yet.

Scope this on a call

How this is calculated. Annual saving = hours per week × people × 52 weeks × hourly cost × 70%. The 70% is deliberate: exceptions, approvals and edge cases still need a person, so we assume an automation removes about 70% of the manual hours rather than all of them. Payback = package price ÷ monthly saving. Indian rupee figures convert at ₹88 to the US dollar. These are estimates to size the opportunity, not a quote — we put a fixed price in writing before any work starts.

Or go straight to the build.

Five ways to work with us. Every engagement is scoped, priced and dated up front — no rolling retainers you can't get out of.

All services →

01

AI automation & agentic workflows

Your team spends hours a week on work a workflow should be doing.

From ₹1.5L / $1,800· 1–3 weeks

02

AI product builds

You have an AI idea that keeps dying in the pilot phase.

From ₹8L / $10k· 4–8 weeks

03

Product engineering

The AI works, but there's no product around it.

From ₹5L / $6k· 3–6 weeks

04

AI strategy sprints

You don't yet know which AI project is worth funding.

₹3L / $3.5k· 2 weeks

05

Forward deployed engineers

You need senior AI capacity without a 6-month hiring cycle.

From ₹6L/mo / $7k/mo· Per quarter

Products we build and run.

All products →

Flagship · Clinical documentation AI

Prexoscribe

Turn the clinical conversation into structured notes, so clinicians spend their time on patients instead of paperwork.

See the product Visit Prexoscribe

  • Ambient capture

    Records the consultation and transcribes it, so nothing has to be typed during the visit.

  • Structured clinical notes

    Transcript becomes a formatted note in the structure the practice already uses.

  • Human-in-the-loop review

    Every generated note is proposed, never auto-filed. The clinician edits and signs off before it lands in the record.

  • Built for real practice workflows

    Designed around how clinics actually run — not a demo that assumes an empty inbox.

Engineering teams · Colleges · L&D

AI training run on live production projects.

Trainees join an active Susam Labs build — AI agents, n8n automation, voice agents, RAG systems, full-stack AI products — and ship real features under senior mentorship. They leave with a portfolio piece, not a certificate of attendance.

Explore programs Request a curriculum
Duration
8–12 weeks, full-time or part-time
Mode
Remote, daily standups, weekly demos
Mentorship
1:1 with engineers who ship AI products
Outcome
Production portfolio project + reference

How we work.

  1. 01 — Week 0

    Call and scope

    Thirty minutes to find out if we're the right shop. Then a one-page spec with a fixed price, a fixed timeline, and a launch date on the calendar.

  2. 02 — Weeks 1–8

    Fixed-scope build

    One outcome, one price, one deadline. Weekly demos, a shared Slack, and real code in your repo from week one.

  3. 03 — Ongoing

    Launch and hand over

    We launch, we hand over, we stay reachable for 30 days at no cost. You own all code and infrastructure — no lock-in.

Proof, not adjectives.

All work →

AI-assisted clinical documentation, with a clinician approving every note

Kids Care Wellness · Prexoscribe

Product engineering and GenAI integration for a European industrial-automation team

Kepler-Labs · Product engineering

Inbound leads enriched, scored and routed to an owner with no human step

n8n · Lead operations

Yesterday's numbers delivered into WhatsApp as a written summary, on a schedule

OpenClaw · Reporting agent

Latest insights.

All posts →

n8n vs zapier vs make

n8n vs Zapier vs Make: Which Automation Platform Should Small Teams Build On?

For small teams evaluating n8n vs Zapier vs Make, n8n is the clear choice for scalable workflow automation because it runs on infrastructure you own, eliminates per-task pricing meters, and avoids vendor lock-in.

6 min read

n8n

What Does It Actually Cost to Automate a Back-Office Process with n8n?

Automating a back-office process with n8n costs between $1,800 and $4,200 in fixed implementation fees, with delivery in 1 to 3 weeks and zero per-task software fees because workflows run on your own self-hosted infrastructure.

6 min read

Common questions.

What does Susam Labs do?

Susam Labs is an AI product studio. We design, build and ship AI-native software end to end — AI agents, voice agents, retrieval-augmented generation (RAG) systems and agentic workflow automation on n8n — and we run hands-on AI training for engineering teams and institutions.

What is the cheapest way to start with Susam Labs?

A productised automation. Back-office task automation on n8n starts at $1,800 and ships in one to two weeks; a WhatsApp reporting agent starts at $2,500 and ships in about two weeks. Each is one named outcome at one fixed price, and most clients extend it once it has paid for itself.

How long does an AI product build take?

Most builds ship in four to eight weeks. Strategy sprints take two weeks. Automations ship in one to three weeks. Every engagement is scoped with a fixed price and a launch date agreed before work starts; if we don't think it can ship in the window, we say so before you commit.

How much does it cost to build an AI product?

Susam Labs publishes fixed prices. Automations start at $1,800. A two-week AI strategy sprint is ₹3L / $3,500. A production AI product build starts at ₹8L / $10,000 and runs four to eight weeks. A forward deployed engineer starts at ₹6L / $7,000 per month.

Who owns the code you write?

You do. Full source, full infrastructure access, and a complete handover on delivery. n8n workflows land in your own instance and OpenClaw agents run on your own infrastructure. There is no lock-in and no hosting dependency on us.

Have a shape of an idea? That's enough to start.

Thirty minutes with us and you'll leave with a concrete opinion on whether to build it, how much it costs, and how fast it can ship.

Book an intro call

Talk for 30–90 seconds. We'll turn it into a structured brief you can edit before sending.