Build AI Applications with Company Data
Build an assistant that answers from your documents—and test whether it works.
Connect a useful AI application to a document collection. Build retrieval, source references, evaluation, and data-access boundaries into the workflow.
A good fit for
Who’s this for?
Developers, data engineers, and technical product teams building applications that use AI at runtime.
Before we start
Working Python knowledge, basic APIs, and Git. Standard exercises use prepared, non-sensitive documents and a lightweight Python reference application.
The programme
Build the skill, step by step.
6 three-hour sessions over 2 weeks remotely. The day-by-day outline below shows what we cover.
Day 1Build the application
Explore LLM APIs, structured outputs, prompt design, document preparation, embeddings, retrieval, and answers with source references.
Day 2Improve reliability
Practise retrieval-augmented generation, metadata filtering, access boundaries, prompt-injection risks, missing-information handling, and human escalation.
Day 3Evaluate and operate
Create an evaluation dataset, test retrieval and answer quality, detect regressions, inspect logs, measure latency and cost, and demonstrate the final application.
What you take away
- Working document assistant
- Ingestion and retrieval example
- Evaluation dataset
- Regression checks
- Operational checklist
The practical details
A clear scope. A team price.
- One-hour preparation call to align goals and check readiness.
- Participant setup checklist, exercise materials, and reusable templates.
- Live, instructor-led training for your private team, delivered in English.
- One 90-minute follow-up clinic within 30 days of the course.
Each teaching day is six live hours, split into two three-hour sessions remotely. Two-day courses run over one week; three-to-five-day courses run over two weeks. On-site delivery uses full days, with travel quoted separately. Dates by arrangement.
Custom implementation, extensive adaptation, extra participants, and travel are quoted separately. Tool subscriptions, cloud usage, and developer accounts are paid separately.
All prices are per private team, excluding VAT. English-language delivery. Dates by arrangement.
Good questions
Before you book
How is this different from the production workshop?
Here you build an application that uses an LLM at runtime. The production workshop focuses on coding assistants, testing, infrastructure, and releasing software. They complement each other.
Do we need to share confidential documents?
No. The standard course uses prepared, non-sensitive documents. Real company documents require separate scoping of access and data handling.
Is AWS required?
AWS is the default cloud context. Azure or GCP adaptations are available by arrangement; we agree the environment during preparation.
What does the starting price include?
The standard remote package for up to 8 people, a one-hour preparation call, readiness checklist, exercise materials, reusable templates, and one 90-minute follow-up clinic within 30 days. All prices exclude VAT. Custom implementation, extensive adaptation, extra participants, and travel are quoted separately. Tool subscriptions, cloud usage, and developer accounts are paid separately.
Can we arrange on-site training?
Yes. On-site delivery uses full teaching days, with travel quoted separately. Remote delivery splits each teaching day into two three-hour sessions. All teaching is in English, with dates arranged by enquiry.
Let’s make it practical
Let’s make this useful for your team.
Tell us about your team, your goals, and your preferred dates. We’ll shape a clear scope together.
How it works
- Your team and goals
Team size, experience, and what you want to achieve.
- A clear scope
We agree the focus, format, and any adaptations.
- Your dates
Remote or on-site, scheduled by arrangement.