AI
AI & Automation
I design AI-assisted workflows and automation layers that connect existing tools, clean up manual handoffs, and make day-to-day operations faster without forcing teams into an oversized platform rebuild.
Service inclusions
- AI assistant workflow design
- Internal tool and API integrations
- Process automation for repetitive tasks
- Reporting, routing, and handoff automation
- Engagement
- 2â6 weeks per engagement
- Investment
- Project-based, typically $1.5kâ$10k
Process
How the engagement runs
A short, written loop with a clear end-of-engagement artefact at every step.
- 01
Workflow audit
We sit with the people who actually do the work and list the top five repetitive tasks. I write them up as a short list of candidate automations.
- 02
Automation design
A one-pager per chosen workflow: trigger, inputs, decision points, the AI or integration layer, and what success looks like for the human at the end.
- 03
Build & test
The automation is built in your own environment, tested against real samples, and rolled out behind a flag so your team can adopt at their own pace.
- 04
Measure & iterate
Two weeks of post-launch observation, a written summary of what actually changed for the team, and a backlog of follow-up automations.
Deliverables
What you walk away with
Concrete artefacts in your own repository, not slides that age out of relevance.
- Working automation in your environment
- A short written playbook for the team that will own it
- A list of edge cases the automation does not yet cover
- A backlog of follow-up automations ranked by impact
Fit check
Is this the right engagement for you?
Good fit
- You have a small team doing too much manual handoff work
- You already pay for tools whose APIs you have never used
- You want a measurable reduction in repetitive operational tasks
- You prefer a thin layer of AI over a full platform replacement
Probably not a fit
- You want a custom-trained foundation model on your own data
- You need 24/7 production monitoring for an AI system
- The task is too vaguely defined to know what "done" looks like
FAQ
Common questions
Which AI providers do you usually work with?
OpenAI and Anthropic for hosted models, Ollama for local inference, and Hugging Face for open-source options. The choice is driven by the data and the latency/cost tradeoffs you can accept.
Can you integrate with our existing tools?
Yes. Most modern SaaS tools have a public API or a webhook, and the ones that do not usually have a Zapier/Make/n8n bridge that is good enough for a first pass.
How do you handle hallucination risk?
For every workflow we agree on what the AI is allowed to do autonomously and what needs a human in the loop. The deliverable always includes the human-review step where it is required.
Do you build chatbots?
I build chatbots only when they are the right shape for the problem. Most of my recent work is backend automation with a chat interface, not a customer-facing chat widget.
Ready when you are
Let's map your engagement
Share a short brief by email or WhatsApp and I will reply within one business day with a proposed scope and a time to talk.