Chatbot development

Chatbot development for support and sales teams

A good support or sales chatbot does three things: it answers from your own content, it knows when it does not know, and it hands the conversation to a person at the right moment. Most chatbots that frustrate customers fail at the second and the third.

Book a scoping callSend us a project brief

Operated by Azinove SAS (France); we work with US teams remotely.

We build chatbots in two ways. For a website, Botsmith, our managed AI chatbot, is the simpler route once it launches: we set it up on your site and documents and run it for you. When the bot has to work inside your product, your internal tools or a complex sales process, we build a custom one.

What a well-built chatbot includes

Whichever route you take, the same pieces decide whether customers trust the bot.

  • Answers grounded in your site, help articles and documents, with a link to the source
  • A clear handoff to your team by email, chat or ticket when the bot should not answer alone
  • Lead capture in sales conversations, sent to the CRM or inbox you already use
  • Transcripts you can review, to see what customers ask and where content is missing
  • Topics the bot must not handle, such as legal, medical or account-specific requests

Botsmith: a managed AI chatbot for websites

Botsmith is the managed option we are preparing. We train it on your website and documents, set up lead capture and human handoff, and look after it for you. It is not in production yet; if you want it on your site, talk to us and we will tell you where it stands.

Custom chatbot development

Some bots need more than a website widget: account data behind a login, actions such as bookings or order changes, or a place inside a mobile app or an internal tool. Those are custom builds, using the same retrieval, evaluation and handoff practices as our RAG and agent work.

Tested before customers see it

Before launch we run the bot on real questions from your inbox or chat history, hostile and off-topic ones included, and check that it refuses, hands off or answers as intended. The tests run again whenever the content or the model changes.

Questions

What clients ask about chatbot development.

Another question? Email hello@cleverai.ai

Do you build chatbots for websites, apps or internal tools?

All three. For a website, Botsmith, our managed chatbot, will often be enough once it launches. Apps and internal tools usually need a custom build.

Is Botsmith available now?

Not yet. Botsmith is a managed AI chatbot for websites that is still in development. Contact us if you want it for your site; we will tell you where it stands and what a custom alternative would involve.

Will the chatbot make up answers?

We design against it: answers come from your content with their sources, and the bot is instructed and tested to say it does not know, and hand off, when nothing relevant is found.

Can the chatbot pass a conversation to a person?

Yes. We set the handoff rules, such as a request for a human, a sales enquiry or a topic the bot must not handle, and route the conversation to your team on the channel you use.

Can a chatbot capture sales leads?

Yes. The bot can ask for contact details at the right point in a sales conversation and send them, with the transcript, to your CRM or inbox.

What content can the chatbot learn from?

Your website pages, help articles and documents such as PDFs, and, for custom builds, structured data through an API. The content is kept in sync so answers change when your pages do.

How is chatbot development different from your RAG work?

A customer chatbot is one application of RAG, with extra work on tone, handoff, lead capture and abuse handling. Internal knowledge assistants follow the same retrieval principles with stricter permissions.

Related

Services, products and reading related to chatbot development.

Talk to the team

Tell us about the chatbot development project you have in mind.

Operated by Azinove SAS (France); we work with US teams remotely.

Book a scoping callSend a project brief