CyberFreezeDev
Loading
Hands-on AI workshops built on your own documents, tools and data rules.
Exercises come from a handful of your team's real tasks, masked where sensitive, so people keep using what they practised.
Most teams already have access to an AI assistant. Few use it for more than rewriting an email, and some paste in things they should not. The gap is not the tool, it is knowing which of their own tasks it handles well, how to check its output, and where the company's line on data sits.
We close that gap with hands-on sessions built around the work each team does. Before the workshop we collect a handful of real tasks from the team: an invoice query, a sales follow-up, a monthly report, a policy question. We mask anything sensitive, and those tasks become the exercises. People practise on work they recognise, so they keep using it afterwards.
The sessions are run by the same engineers who build AI workflow automation and private AI systems for clients, so questions about what is possible get a real answer, not a slide.
| Format | Who it is for | What happens | People leave with |
|---|---|---|---|
| Leadership briefing | Owners, partners, department heads | What AI does well and badly today, costs, data risk, how to choose first projects | A shortlist of candidate use cases and a view on policy |
| Half day hands-on | Any non-technical team | Exercises on the team's own tasks in the assistant they use, plus checking and correcting output | A saved prompt library for their tasks and the data rules in one page |
| Full day build session | Operations and technical staff | Build a working automation in n8n, or a document question and answer setup, from start to finish | A working workflow they built, and the notes to build the next one |
| Follow-up clinic | Anyone who attended | A session a few weeks later on what people tried, what broke and what to do next | Fixes for real problems found in daily use |
Ledgers, GST and TDS replies and bank matching, with a person checking every figure.
Summarising ledgers and notices, drafting replies to GST or TDS queries for a person to verify, reconciling bank lines against invoices, and reading invoices and bills into structured rows. Always with the rule that a person checks every figure.
Prospect research, follow-ups in the customer's language, CRM call notes and WhatsApp qualification.
Researching a prospect from public information, drafting follow-ups in the customer's language, summarising call notes into the CRM, and qualifying WhatsApp enquiries.
SOPs turned into checklists, purchase order data extracted and repetitive steps spotted for automation.
Turning SOPs into checklists, extracting data from purchase orders and delivery notes, and spotting which of their own repetitive steps could become an automation.
Briefs, ad copy, content calendars and reports, within clear limits on claims and voice.
Campaign briefs, ad copy variations, content calendars and reading performance reports, with clear limits on claims and brand voice.
Model APIs, retrieval over documents, local models with Ollama, n8n and evaluation sets.
Calling model APIs, building retrieval over company documents, running a quantized model locally with Ollama or llama.cpp, connecting tools through n8n, and testing output quality with an evaluation set.
Every session covers what must not be pasted, which tools are approved, how to check output and when to escalate.
Every workshop covers the same short set of rules, adapted to your policies:
If you do not yet have an AI use policy, we leave you with a one page draft based on what we learned in the sessions, for your management and legal adviser to finalise. For the larger question of which AI projects to fund, see AI strategy consulting.
Who attends, which tools they use, the data rules and what success looks like.
Who attends, which tools they have, what the company's data rules are, and what a good outcome looks like.
Three to five masked examples from each team become the raw material.
Three to five real examples from each team, with sensitive details masked by you or by us under an agreement.
Each task becomes a guided exercise with a worked answer and common mistakes.
Each task turned into a guided exercise with a worked answer and the common mistakes to look for.
Mostly hands-on, each person on their own laptop in their everyday tools.
Mostly hands-on, with each person working on their own laptop in the tools they will use afterwards.
A prompt library, exercise files, the data rules page and an optional recording.
Prompt library, exercise files, the data rules page and the recording if you want one.
No, except for the build session. The hands-on workshops assume people can use email and a spreadsheet. Everything else is shown step by step.
The one your company already uses or plans to approve. We do not push a particular subscription. If you have none, we explain the trade-offs between the main options for your data and budget.
Yes, remotely for teams in India, the UK, the Gulf, Singapore and elsewhere. In-person sessions outside Chennai can be arranged case by case.
Yes. The technical session can include running an open-weight model on a laptop or workstation, which is a good way for an IT team to see what a private deployment involves before investing in one.
Last updated