I Built an AI Employee, Then Had Him Make His Own 2-Minute Explainer

Jake — Wed, 22 Jul 2026 06:28:32 +0000

I built an AI employee named Jack — then had him make his own two-minute explainer video.

That sounds like a gimmick, but the real point is more practical: most businesses do not need another chatbot. They need something that can take repetitive computer work off the team's plate.

What I mean by "AI employee"

I do not mean a floating chat widget that answers FAQs.

The model I am building is closer to a digital teammate with:

its own computer / execution environment
its own email and phone number
access to approved business tools
a company knowledge base
operating procedures and personality/context files
the ability to create docs, update systems, summarize work, and run workflows

The interface can be Telegram, email, Slack, WhatsApp, or a dashboard. The interface is not the product. The product is delegation.

The problem: work about work

A lot of teams spend a shocking amount of their day feeding the machine:

CRM updates
scheduling
invoice follow-up
lead follow-up
customer intake
copying information between systems
weekly reports
internal status updates
turning messy meeting notes into action items

That is the work I think AI employees are best suited for first.

Not replacing the human relationship layer. Replacing the repetitive computer layer underneath it.

A practical example

Imagine a local service business gets an inbound lead:

"Hey, I need help next week. We are in Tampa. Can someone call me?"

A basic chatbot might answer the customer.

An AI employee could instead:

read the inquiry
extract the service type, location, urgency, and missing details
create or update the CRM record
draft a human-approved reply
schedule a follow-up task
notify the owner if the lead looks high-value
make sure the lead does not go cold

That is much more useful than another generic AI answer box.

Human approval still matters

The best version is not fully reckless automation on day one.

The sane workflow is:

AI drafts
human approves
system logs what happened
edge cases improve the SOP
more trust is earned over time

That is how you move from "AI toy" to actual business infrastructure.

Why I made the video

The video is short on purpose. I wanted to explain the concept in two minutes, in plain English, and show the meta point at the same time:

An AI employee should be able to help create the sales, marketing, ops, and admin work around itself.

Here it is again if you want to watch it:

https://youtu.be/eURIynQGWCI

The real question

The question I keep asking business owners is simple:

What is one repetitive computer task your team does every week that you wish just happened correctly in the background?

That is usually where the first AI employee workflow should start.

Machines handle the repetitive work. People handle the people.

Local-first RAG for privileged legal documents: why citations need verification

Jake — Tue, 14 Jul 2026 01:47:12 +0000

Lawyers want the same thing everyone else wants from AI: ask a question, get a useful answer, move faster.

The difference is that legal documents often come with privilege, confidentiality, protective orders, client expectations, and professional responsibility duties. Uploading a contract set, deposition transcript, pleading file, or discovery bundle into a generic cloud chatbot is not a neutral implementation detail. A vendor sits in the data path.

That is the design problem behind DocuChat: an open-source, local-first document chat app for attorneys.

The core idea is simple:

Put the AI next to the documents, not the documents next to the AI vendor.

The architecture

DocuChat is a self-hosted legal document RAG stack:

Local models through Ollama
FastAPI backend bound to loopback
LanceDB embedded vector store
PyMuPDF / Docling / Tesseract for PDFs, tables, and OCR
Matter-scoped retrieval so one client/matter cannot bleed into another
Verified page/span citations before answers are displayed

The query path does not require OpenAI, Anthropic, Google, or any cloud AI API key. After local models are downloaded, the workflow can run offline.

Why ordinary citations are not enough

A citation-looking string is easy for an LLM to generate. That does not mean the cited text actually supports the claim.

DocuChat treats citations as a mechanical verification problem. The model can propose an answer and source quote, but the app checks whether that quote actually appears in the retrieved source text at the claimed page/span. If a claim cannot be verified against the source, it is dropped instead of shown.

That creates a more useful failure mode for legal work:

Better: “I could not find that in the documents.”
Worse: a confident answer with a fake or sloppy citation.

What this is not

DocuChat is not an AI lawyer. It does not give legal advice, draw legal conclusions, or act outside the user’s machine.

It is a cited-retrieval assistant for document-heavy work: contracts, transcripts, scanned PDFs, exhibits, pleadings, and matter files.

Why open source matters here

For sensitive workflows, “trust us” is not enough. Legal technologists should be able to inspect the data path, citation logic, storage layer, and model boundaries.

The repo is here:

https://github.com/janderswag/docuchat.app

If you care about local-first AI for private legal documents, a GitHub star/watch helps validate that this is worth building further.

DEV Community: Jake