Every product starts as a thought.
You bring the problem. We ship the system that solves it.
IN MIND IT is an AI lab with eight years of building software for large companies. We build AI agents, RAG, computer vision and decision systems, from a first sketch to production. Here is how an idea gets there.
We give it a mind.
A model that knows your business, not just the internet.
Move the cursor across the brain to fire a thoughtMachine learning, deep learning and generative AI. We pick the right model for the job, from open-source to frontier LLMs, and fine-tune it with LoRA on your data when off-the-shelf is not enough.
We teach it to see.
Cameras that count, check and flag problems on their own.
Computer vision that detects, tracks and counts in real time. YOLO models and OpenCV pipelines running on cameras, drones, edge devices and conveyor lines.
Then it learns to work as a team.
Your team asks. The agents answer with sources and do the follow-up work.
Point at an agent to pass it a messageAgents that search, fill in and check things for your team. They show where every answer came from and ask a person when they are not sure. Multi-agent systems on LangGraph and CrewAI, connected through MCP and running on Amazon Bedrock AgentCore.
We teach it to decide.
Decisions backed by data and risk limits, not gut feeling.
Markets move like dominoes: one event tips the next. Our models learn to read that chain, weigh each move against its risk and choose the one worth taking, inside trading systems we build or the ones you already run.
We build it a body.
Less manual work. Forms, reports and handovers run on their own.
AI automation for the steps your team repeats every day. A person steps in only for the exceptions. Workflows in n8n and AWS Step Functions, APIs in FastAPI and Node.js, firmware in C/C++.
We send it out into the world.
Systems that keep working after launch day, with us still on call.
Tested, put into production and watched. Eight years of building and running systems for large companies.
Every system brings the singularity closer.
In practice: less routine for your people, more time for the work only they can do.
Machines that learn, see, reason and act together. We build them today, one working system at a time.
Building your future technologies in the present.
What we build
Six things
we ship.
Pick a service to read about it, or watch its scene from the film.
ML & generative AI
A model that knows your business, not just the internet.
Computer vision
Cameras that count, check and flag problems on their own.
AI agents & RAG
Your team asks. The agents answer with sources and do the follow-up work.
Trading & decisions
Decisions backed by data and risk limits, not gut feeling.
AI automation
Less manual work. Forms, reports and handovers run on their own.
Launch & support
Systems that keep working after launch day, with us still on call.
A production byIN MIND IT
Crew
- AI automation
- AI agents & multi-agent systems
- RAG & knowledge assistants
- MCP servers & tool use
- LLM fine-tuning (LoRA)
- AI guardrails, evals & observability
- Cloud AI on AWS
- Generative AI
- Machine learning & deep learning
- Chatbots & AI assistants
- Computer vision · YOLO, OpenCV
- Trading systems
- AI decision-making & risk
- Applied maths & data analysis
- System architecture & design
- Web, mobile & desktop
- APIs & integrations · REST, SOAP, TCP
- IoT & sensor networks
- Robotics, drones & aerial scouts
- Industrial automation & conveyor systems
- Blockchain & smart contracts
- Migration & refactoring
- DevOps, CI & testing
- UX/UI, branding & campaigns
Starring
- Python
- Node.js
- C/C++
- LangGraph
- LangChain
- CrewAI
- LlamaIndex
- MCP
- Amazon Bedrock
- Bedrock AgentCore
- Step Functions
- AWS Lambda
- OpenSearch
- OpenAI
- Anthropic
- Hugging Face
- YOLO
- OpenCV
- PyTorch
- TensorFlow
- scikit-learn
- FAISS
- Weaviate
- pgvector
- FastAPI
- n8n
- React
- Next.js
- Docker
- Kubernetes
- Terraform
- Google Cloud
Filmed in Bulgaria
Behind the scenes
AI you can
put in production.
A demo that works once is easy. We build the parts that keep it working: checks on what goes in, checks on what comes out, and numbers on everything in between.
Guardrails
- Prompt-injection and role-hijack detection
- System-prompt leak protection
- PII redaction and unsafe-request filters
- Validated, structured outputs
Evals
- Golden datasets and held-out tests
- LLM-as-a-judge with human review
- Precision, recall and F1 on every release
- Prompt and schema versioning
Reliability
- LLM routing and model fallbacks
- Outputs checked against allowed values, repaired or retried
- Context and conversation memory
- Fail-closed safety gates
Observability
- Tokens, cost and latency per call
- Budget caps per feature
- Feedback loops from real users
- Drift and discrepancy monitoring
Final reel · your project
Write the
next scene.
Pick what you need. We turn it into a brief you send in one click.
Directed by
Yordan
Daskalov
CEO · Co-founder
Leads the company and its clients, and still reviews the architecture.
Kristiyan
Dimitrov
CTO · Co-founder
Leads the technology and the AI work, and talks to every client directly.
Two founders on every project. The people you pitch to are the people who build it.