/AI-native: AWS + Claude
AI training for teams that are going to use it for real
Hands-on workshops so your team uses AI in everyday work and takes it to production without opening security holes. We work on your code, your AWS account and your real use cases, not toy examples.
/Diagnosis
Signs it is time
Everyone uses AI their own way
Some engineers already code with agents and others haven't tried them. There are no shared criteria for what to use, when, and with which data.
Security blocks it or never hears about it
Either AI is banned and people use it anyway on the side, or it gets used without anyone thinking about credentials, customer data and prompt injection.
You want to build an agent and don't know how
The business wants an assistant grounded in company data, and the team isn't sure where to start with RAG, tools, evaluations and cost.
Licenses nobody makes the most of
You bought AI tools, but real usage is low because nobody showed how they fit into the team's workflow.
/What we deliver
Concrete deliverables, in your repository
Claude Code and coding agents
How to give the agent context with CLAUDE.md, which permissions to grant, how to review what it produces and which tasks it truly handles well.
Agents and RAG on Amazon Bedrock
From the first model call to an agent working with your documents and tools, with evaluations before every change.
MCP servers
How to connect AI to tickets, databases or internal APIs through MCP, and which controls to put in place before opening that access.
AI security
Prompt injection, data leakage, overprivileged agents and third-party MCP servers: how they get attacked and how to defend them.
AI costs
How to read token usage, pick the right model for each task and set alarms before the bill arrives.
Material to keep practicing
Exercises, sample configurations and written guides that stay in your repository for the team and for anyone who joins later.
/How we work
Our process
Discovery
A conversation with the team to understand its stack, its experience with AI, its security policies and what it wants to achieve.
Tailored program
We choose the topics and build the exercises on your repositories and AWS account, or on a sandbox environment if you prefer.
Workshops
Hands-on sessions, remote or in person, where everyone works with their own tools from the very first minute.
Follow-up
After the workshops we check how the team is using what it learned, answer questions and adjust the configurations that came up.
AI training your team puts to work on real projects
Most generative AI courses teach with generic examples that don't fit real work afterward. We build every workshop on the team's own stack: its repositories, its pipelines, its AWS account and the problems it has today. What gets practiced in the session can be used on the job right away.
We are engineers who use these tools every day to run infrastructure on AWS, not instructors who saw them in a demo. That is why we also show where AI fails, which tasks you should not hand off to it and how to review its output before it reaches production.
AI security: what your team needs to know
An agent that reads documents, emails or web pages can pick up instructions hidden in that content. This is known as prompt injection, and it is the most underestimated AI risk in companies. If the agent also has broad permissions or uncontrolled access to MCP servers, a single malicious instruction is enough to leak data or trigger actions.
In the workshops we demonstrate these attacks live and show how to defend against them: least-privilege permissions per tool, separation between data and instructions, guardrails for personal data, human approval for irreversible actions and logging of every call. The goal is for the development and security teams to speak the same language.
From curiosity to your first agent in production
For teams that want to build with AI, the path runs from the first model call on Amazon Bedrock to an agent with RAG over company documents, scoped tools and an evaluation suite that runs before every change. Every step is done with code that stays in the team's repository.
We also cover cost: how each token is billed, which model fits each task, when to use caching and how to set alarms per team or per feature. That way, the first agent reaches production with clear numbers and no surprises on the bill.
/FAQ
AI training: frequently asked questions
Another question? Write to hello@nubent.com and an engineer will answer.
Who is the AI training for?
For engineering teams: development, platform, DevOps, data and security. We adapt the content to the team's level, from people who don't use AI at work yet to those who already want to build agents. We also run short sessions for technical leaders who need to decide how to adopt it.
Are the workshops remote or in person?
Both. Remote sessions work well for distributed teams and can be split into short blocks. In-person sessions are better for working more intensively over one or two days. Either way, everyone practices with their own tools and their own code.
Do you work with the company's real code?
Yes, if your security team approves it, because that is where the training pays off most. If not, we set up a sandbox with a repository and a lab AWS account that mirror your stack. We never use production credentials during the workshops.
What is prompt injection?
It is an attack in which someone hides instructions inside content an AI model processes, such as a document, an email or a web page, to make the agent do something it shouldn't. It is mitigated with least-privilege permissions, by separating data from instructions, by filtering inputs and outputs, and by requiring human approval for sensitive actions.
Related services
Claude Code for teams
Claude Code adoption for engineering teams: setup, CLAUDE.md, permissions, MCP, CI/CD, code review, credential security and measuring the impact on delivery.
Learn moreAI Agents with Bedrock
Agents that carry out real tasks with Amazon Bedrock and AgentCore: tool use, MCP, Knowledge Bases, multi-agent orchestration, identity and observability.
Learn moreMCP servers
Custom MCP servers that connect Claude and other agents to your internal systems and AWS, with authentication, least privilege, audit logging and deployment in your account.
Learn moreSecurity and compliance
AWS security: access, network, encryption and secrets fixed, with automated controls and continuous evidence for SOC 2, ISO 27001, PCI DSS, HIPAA or Spain's National Security Framework (ENS).
Learn moreNext step
Let's talk about your AI training project
30 minutes with an engineer. Concrete next steps, whether or not you work with us.