Automation & AI agents
AI chatbot and support assistant development
Chat assistants trained on your own content that answer customers accurately, hand off to a person when needed, and never make up a policy.
- Starting from
- $2,500
- Plus model usage fees, billed to your own account.
- Typical timeline
- 1–3 weeks
A good support assistant answers from your documentation, help articles and policies — not from the open internet — and says so when it does not know. A bad one invents answers, which is worse than no chatbot at all.
We build assistants for websites and apps that are grounded in your own content, cite where an answer came from, hand the conversation to a person when they should, and log every conversation so you can see what customers are really asking.
We already ship one. Focusara, our own app, has an AI coach you can ask questions in plain language: "why did my focus dip this week?" It answers from the user's real sleep, mood, hydration and habit data, and when a metric wasn't tracked it says so instead of inventing a number. It runs Gemma 3 entirely on the phone, with no cloud and no per-message cost, which is an option for assistants that handle private data.
What you get
- Chat assistant for your website or app
- Answers grounded in your documents and help content
- Hand-off to your team by email or helpdesk
- Conversation logs and common-question reports
- Guard rails against off-topic and unsafe answers
- Cost limits and usage reporting
How we work
- 1
Free scoping call
We talk through what you are building, who it is for and what has to be true on day one. You leave with a recommended approach and a budget range.
- 2
Gather your content
We collect the documents and answers the assistant is allowed to use.
- 3
Test against real questions
We measure answers against questions your customers actually ask, before launch.
- 4
Launch and improve
Monthly review of conversations to close gaps in the content.
Work we have shipped
Technology
- OpenAI API
- Azure OpenAI
- Retrieval-augmented generation
- Next.js
- Node.js
- Gemma
- On-device AI
Frequently asked questions
Will it make things up?
It is built to answer only from your content and to say when it does not know. We test it against real customer questions before launch and review conversations afterwards.
Can it hand off to a person?
Yes — by email, your helpdesk or live chat, with the conversation so far attached, so the customer does not repeat themselves.
What does it cost to run?
Usually a small amount per conversation in model fees. We set limits and show you usage, so the cost stays predictable.
Let's talk about your project
Book a free call. You will come away with a clear approach, a realistic timeline and a budget range — whether or not you work with us.
Related services
Business Process Automation
Scheduled jobs, reminders, payouts and reports that run your operations without anyone watching — like the 28 jobs behind a staffing marketplace.
AI Agents & LLM Workflows
AI that does work, not just chat — turning speech, documents and messages into structured data and actions in the systems you already use.