bas@headingfwd: ~/ai-engineering — zsh
HeadingFWD
bas@headingfwd:~$ ./hello --who

HeadingFWD

AI engineering & consultancy · Bas Wenneker — AI Lead / Engineer
I help teams get real value from Generative AI — designing and building agents, assistants and AI workflows that actually make it to production, training dev teams, and consulting on AI strategy.
// specialities
*Agentic workflow development
*AI strategy & consulting
*Agentic coding training for dev teams
*AI techniques: RAG, graphs, memory and more
tip: type for commands · to browse my work fullscreen · or just ask
bas@headingfwd ~$
onlinemainutf-8Plaintext version for agents (llms.txt)llms.txt18 lines · /help
bas@headingfwd:~/portfolio
// selected work — AI engineering & product design
→ or view my LinkedIn profile
011 / 7

AI Writing Assistant

AI writing assistant that guards the house style — data stays in-house
Government · live · Initiator / AI engineer
LLMWritingMarketingPythonVectorDB

In short

An AI-powered writing assistant for the editorial team of a large public-sector organization. The tool rewrites text into a clear, error-free message that meets the organization's style guide, word lists, style rules and accessibility requirements (B1 level). One consistent house style for the whole editorial team — while sensitive data stays inside the organization's own environment.

Problem

At an organization with one of the most-visited websites in the Netherlands, which sends millions of letters a year, writing is a core task. Editors have to deliver text that meets a style guide, word lists, style rules and accessibility requirements. Generic AI tools (ChatGPT, Copilot) fall short for this:

  • No knowledge of the house style or tone of voice
  • Generic suggestions without the organization's context
  • Inconsistent output across different prompts
  • Privacy concerns with sensitive company and personal data

Approach

I helped shape a custom AI writing assistant from the ground up. We worked in close co-creation with members of the editorial team. Early on there was suspicion and resistance; as we collaborated more closely and delivered results, trust and adoption grew.

The solution runs inside the organization's own environment (an internal URL like schrijfhulp.intranet.nl), so even commercially sensitive and personal data never leaves the organization.

How it works

The user pastes a text; the assistant analyzes it and returns a rewritten version per sentence, with remarks. Examples from the demo — a Dutch-language tool (original → rewritten):

  • "Door de afgelopen week ben ik beezig geweest met het ontwikelen van een nieuwe AI-gestuerde schrijfhulp tool." → "Vorige week werkte ik aan het maken van een nieuwe AI-gestuurde schrijfhulptool." (simpler, B1, spelling fixes, compound word)
  • "…zodat je profesioneler overkomt in je communicatie." → split into two sentences, "professioneler" corrected (clarity + spelling)
  • "Vervolgens geeft hij suggesties…" → "Vervolgens geeft de tool suggesties…" (clearer reference)

The output appears as a table with three columns: Original sentence · Rewritten sentence · Remarks.

Benefits:

Benefit Explanation
🔒 Data security Also for commercially sensitive and personal data; everything stays in your own environment
⚡ Efficiency Instant result, ready while you wait
🎯 Consistency A style guide is interpreted differently by everyone; AI does it consistently
📚 Word lists Enforce jargon — or deliberately avoid it
👥 B1 level Write so the average reader can follow it easily
✨ Everyone can do it Turns every employee into a good writer

"By constantly getting new suggestions, it helps me in the creative process and it instantly meets the writing rules we follow!" — Editor

Tech & stack

  • ☁️ Azure OpenAI LLMs — LLM API provider
  • 🐍 Python — backend
  • 🤖 Agentic programming — AI architecture pattern
  • 💻 VSCode — IDE
  • 🔎 Vector database — surfaces the style guide and word lists for the tool

Status

Live — custom client project at a large public-sector organization.

work with me →