AI chatbot development cost
A chatbot pilot over your documents costs $5,000 to $10,000.
An AI chatbot pilot costs $5,000 to $10,000 with us. It answers from one set of your documents, shows the source for each answer, and comes with an evaluation set that measures how often it is right. A wider rollout after the pilot is scoped after a call. Language model usage is billed by the provider to your account, separately from our price.
Updated · Prices in USD · Acasa Labs
The numbers
What an AI chatbot costs with us.
| What you get | Price (USD) | Timeline | What it covers |
|---|---|---|---|
| Chatbot pilot over one document set | Price (USD)$5,000 to $10,000 | TimelineAgreed in the scope | Search or chat over one document set, answers with sources, and an evaluation set that measures whether the answers are right. |
| Production rollout after the pilot | Price (USD)Scoped after a call | TimelineScoped after a call | More document sets, permissions, channels such as WhatsApp or your app, and monitoring. Priced from what the pilot shows. |
| Dedicated developer | Price (USD)$1,500 to $2,000 a month | TimelineMonthly, about 160 hours | One developer full time on your roadmap, in your repository, with a senior reviewing pull requests. |
| Maintenance and support | Price (USD)$100 to $300 a month | TimelineMonthly | Updates, fixes, monitoring, and the small changes that come up after launch. |
We do not put a number on a production chatbot before the pilot, because the pilot is how we find out what the documents need. A dedicated developer at $1,500 to $2,000 a month is one way to run the rollout.
Every price on this page is our own rate card, in US dollars. The minimum project is $1,500. A dedicated developer is $1,500 to $2,000 a month. Hire engagements start with a risk-free first week, and you own the code, repositories, and cloud accounts from day one. Indian clients are invoiced in INR by Acasa Labs (OPC) Private Limited. See every engagement model and rate.
01
What drives the cost of an AI chatbot
The chat window is the cheap part. The cost is in making the answers right, and proving it.
The documents
How many, in what formats, and how clean. Clean PDFs and web pages are quick. Scanned pages, tables, and spreadsheets need OCR and careful parsing before retrieval works.
How right the answers must be
A chatbot that suggests articles can be wrong sometimes. One that quotes policy, prices, or contract terms cannot. Stricter answers need stricter retrieval, citations, and refusing when the source is not there.
The evaluation set
A list of real questions with correct answers, written with your team. It is how the pilot is judged, and it is included in the range.
Who can see what
If different users may only see different documents, permissions have to apply to retrieval, not only to the screen.
Where it lives
A chat page on your site is simplest. Inside your app, in WhatsApp, or wired to your help desk adds integration work.
02
Why we start with a pilot
A chatbot demo that answers five questions well is easy. A chatbot that answers your customers' real questions well is not, and nobody knows how hard until they try it on your documents.
The pilot answers that question for a fixed price: one document set, a working chatbot, and a measured score against your own questions. If the score is good, you roll out with confidence. If it is not, you have spent $5,000 to $10,000 to learn that, instead of far more on a full build.
03
Running costs after launch
A chatbot has a running cost our build price does not cover: language model calls and the vector database, billed by those providers to your accounts. We build in token and cost logging per call, as in SigmaCPM, so the cost per conversation is visible from the first day rather than a surprise on the invoice.
Example builds
AI chat and answers we have built.
These are from our own products. They are larger than a pilot, and they show the parts of chatbot work we have done before.
Grounded chat over a student's own material
NoteTubeNoteTube's chat answers from the PDFs, lectures, and notes a student uploads, stored in Qdrant per user and per session, and streamed back over server-sent events. Users choose between five model providers.
Answers that refuse to guess
SigmaCPMSigmaCPM's AI must quote the tender page for every value it extracts. If the quote is not on that page, the value is dropped, and a second model run re-checks bond terms.
04
How to spend less on an AI chatbot
- 01Start with one document set and one audience, such as support articles for customers.
- 02Write the evaluation questions with the people who answer them today. It shortens the pilot.
- 03Clean the documents first: remove old versions and duplicates.
- 04Use a smaller, cheaper model where the score allows it, and a larger one only where it must.
- 05Put the chatbot on one channel first. Add WhatsApp or the app after the answers are proven.
05
When a cheaper option is the better choice
If your help desk or website platform already offers an AI assistant trained on your help articles, try it first. For a small set of clean, public articles, it may be good enough without a build.
If the questions are few and predictable, a well-organized FAQ page or a scripted bot with buttons may serve customers better than AI.
A custom chatbot is worth building when the answers come from private, messy, or changing documents, when wrong answers are costly, or when it has to work inside your own product.
Rough number first
Pick platforms, features, and integrations, and get a range and a timeline in a minute.
Questions
AI chatbot cost, asked often.
How much does it cost to build an AI chatbot?
With us, a pilot over one document set costs $5,000 to $10,000, including an evaluation set. A production rollout is scoped after a call, based on the pilot.
Does the price include OpenAI or other model costs?
No. Model and vector database usage are billed by the providers to accounts in your name. We make the cost per conversation visible.
Will the chatbot make things up?
We design it to answer from your documents, show the source, and say it does not know when the source is missing. The evaluation set measures how often it gets things right.
How long does a pilot take?
It depends on the documents, so we agree the timeline in the written scope rather than quoting one here.
Can the chatbot work in WhatsApp or our app?
Yes, after the pilot. Each channel is integration work and is priced in the rollout scope.
Who owns the chatbot?
You do: the code, the index, and the accounts for the model and the database.
AI chatbot cost
Get a fixed estimate for your chatbot pilot.
No call needed to start. Tell us which documents it should answer from and who will ask the questions, and we reply with a scope and a number.