02Service
LLM & ChatGPT integration
GPT and Claude inside your product, answering from your data, with guardrails you can explain to legal.
Put a language model in your product without the surprises.
Connecting to an LLM API takes an afternoon. Making it answer from your data, refuse what it should refuse, stay within budget and behave the same next month takes real engineering.
I've built ChatGPT-based assistants for clients since the API became usable for business. The pattern that works: ground answers in your own content, give the model a few well-defined actions, and test every change against questions your users actually ask.
A good fit if
- You want a support or sales assistant that knows your product
- Your team searches through docs, tickets or contracts all day
- You need an AI feature in your SaaS and a clear cost per user
- You've had an LLM make something up in front of a customer
What you get
- 01
Search over your own content
Your docs, tickets and records indexed so answers come with sources, and stay current when the content changes.
- 02
An assistant or feature in your product
Chat, a sidebar, an API or a background job. Whatever fits how your users already work.
- 03
Guardrails and privacy
Server-side permission checks, redaction where needed, and clear rules for what the model is allowed to say and do.
- 04
Cost and quality tracking
Token spend per user and answer quality over time, visible in a dashboard instead of a surprise invoice.
How a project runs
Week 1
Audit
What content exists, who should see what, and the twenty questions users ask most.
Weeks 2 to 3
Prototype on real data
A working version on your actual content, tested with a few internal users.
Weeks 4 to 6
Production
Auth, rate limits, monitoring and the release itself. Then we watch real usage.
Ongoing
Tune
Prompt, retrieval and model changes, each tested against the eval set before it ships.
- OpenAI API
- Claude API
- Gemini
- LangChain
- LlamaIndex
- Embeddings
- PostgreSQL / pgvector
- Langfuse
- Next.js
- Laravel
- Python
Related work
- File 002AI agentsIn beta
Altrun
An AI assistant that turns incoming signals into scheduled actions- Role
- Founder, engineer
- Year
- 2026
- Stack
- LLM tool calling, Agentic workflows, Telegram Bot API
- File 003AI platformLive
Name withheld under NDA
Name withheld under NDA- Role
- Lead engineer
- Year
- 2025
- Stack
- LLM agents, TypeScript, Next.js
- File 004Customer messagingLive
Name withheld under NDA
Name withheld under NDA- Role
- Lead engineer
- Year
- 2024
- Stack
- WhatsApp Business API, Node.js, React
Questions people ask
Is our data used to train the model?
Not through the business APIs I use. Your data is sent to answer a request and isn't used for training by default. I document exactly what leaves your systems.
Can it run on our own servers?
Yes, if privacy rules require it. Open-weight models can run in your cloud account. They cost more to operate and are a bit less capable, so we decide based on what the data actually needs.
How do you stop it making things up?
It answers only from retrieved sources, cites them, and says when it doesn't know. We test for this specifically before launch.
What does it cost to run?
It depends on volume and model choice. You get an estimate per thousand conversations during the audit, before any build starts.
Contact
Tell me what you're trying to fix.
I read every message myself and usually reply the same day. A rough idea is enough to start.
Or send a short briefHours
Your hours work for me.
San Francisco
--:--
New York
--:--
London
--:--
Berlin
--:--
Dubai
--:--
Singapore
--:--
Sydney
--:--
Surat
--:--
Me
I keep flexible hours and move my day to overlap with yours, whether that's a morning standup in New York or a late release in Sydney.