EBOXLAB/ai Book a call

AI practice of EBOXLAB · Conifer, Colorado · since 2004

Private AI for Colorado companiesthat can't afford a wrong answer.

We build AI tools and automation, plan them and train your team on them. On your own servers when the data can't leave the building.

Book a call See the four services
private-ai · your serverillustration

you@your-server:~$ ask "What does our records policy say about retention?"

[search] the files you already have, on your server

[answer] drafted, every claim cites its source

[review] staff can flag errors; sensitive actions wait for a person

[egress] bytes sent to public AI services: 0

private model online · 0 bytes left the building

How a private deployment behaves. An illustration, not a live system.
  • private AI assistants
  • document search with cited sources
  • AI voice agents
  • AI agents inside your systems
  • AI workflows
  • AI pipelines
  • document processing
  • process automation
  • SaaS built with AI
  • readiness assessment
  • SB 26-189 readiness
  • team training
  • on-premise
  • fully offline

[ services ]

Four services.The people who build your system train your team on it.

01/ 04

Custom AI tools

AI built around your documents, your calls and your systems. Private by default.

cloudon-premisefully offline

[ how it starts ]

A focused pilot on your real files. It typically runs a few weeks.

[ what we build ]

  • Private AI assistants on your own documents, with search that cites its sources
  • AI voice agents that answer and log calls
  • AI agents that act inside your systems

[ what you get ]

  • The working tool, running where you chose: your cloud, your servers or a network with no internet
  • Answers that cite their sources, and a way for staff to flag errors
  • A person in the loop for sensitive actions
  • Code, prompts and documentation

[ examples ]

  • Find the policy, clause or record and show the page it came from
  • Answer the phone and log every call
  • Let staff question files that can't leave the building

02/ 04

AI workflows & automation

Software and SaaS built with AI to optimise how your processes run.

[ how it starts ]

One process, mapped end to end, then a pilot on it.

  1. intake
  2. extraction
  3. validation
  4. routing
  5. reporting

[ what we build ]

  • AI workflows and AI pipelines
  • Document processing
  • Process automation and business optimisation
  • Software and SaaS products with AI inside

[ what you get ]

  • The pipeline running in your own accounts
  • Integrations with the systems you already use
  • A person approves where it matters
  • Code and documentation

[ examples ]

  • Read incoming documents, pull out the fields, check them, send them to the right person
  • Turn a report someone assembles by hand into one that builds itself
  • Add an AI step to a tool you already use, with approval before anything goes out

03/ 04

AI consulting

Know where AI pays off, what it takes and what to leave alone, before you spend on it.

[ how it starts ]

An assessment. It takes days.

[ what we do ]

  • AI readiness assessment
  • Use-case shortlist scored by value, effort and risk
  • Roadmap, vendor and model selection
  • Data privacy and governance
  • Colorado SB 26-189 readiness

[ what you get ]

  • A scored shortlist and a roadmap you can act on
  • A recommendation on vendors and models
  • Privacy and governance rules for AI use
  • For SB 26-189: an inventory of where AI influences consequential decisions, consumer notices, human review, records

[ examples ]

  • Which of our AI ideas is worth a pilot?
  • A cloud model, or one on our own server?
  • What do we need in place by January 1, 2027?

04/ 04

AI training for companies

Hands-on, on your team's own workflows, with the tools you already have.

[ how it starts ]

With how your team works today. The sessions are built around it.

[ what we teach ]

  • ChatGPT, Microsoft Copilot, Claude and Gemini
  • The tools we build for you
  • A leadership track
  • A staff track

[ what you get ]

  • Hands-on sessions on your own work, not generic slides
  • A usage policy your team can follow
  • A prompt library for your workflows

[ examples ]

  • Staff already use AI, but nobody has agreed the rules
  • Leadership needs to know what to approve
  • A new AI tool has to land with the team

[ also ]

  • AI security review
  • Fractional AI lead
  • Support and monitoring after launch

[ process ]

Five stages.A fixed scope for each one.

You see what a stage covers before it starts.

  1. 01days

    Assessment

    Your workflows, data and risks. A shortlist of use cases scored by value, effort and risk.

  2. 02a few weeks

    Pilot

    One focused use case, working on your real files.

  3. 03after the pilot

    Build

    The full tool or pipeline, connected to your systems. Cloud, on-premise or fully offline.

  4. 04with your team

    Training

    Hands-on sessions for leadership and staff, led by the people who built the system.

  5. 05after launch

    Support

    Monitoring and support once the system is in daily use.

[ what stays with you ]

  • The codeEverything we write for you.
  • The modelsRunning where you chose to run them.
  • The promptsReadable, editable, yours.
  • The dataIt never stops being yours.
  • The documentationHow it works and how to run it.
  • Your accountsSet up in your name. No lock-in.

[ why eboxlab ]

Built like evidence,not like a demo.

  • Build, consult and train under one roof

    One team plans the work, builds the system and trains your people on it.

  • Private AI

    Models on your own servers, even on networks with no internet. Client, patient and case files never go to a public AI service.

  • Security and forensics DNA

    25 years of digital forensics and data recovery behind the founder: audit trails, evidence-grade logging and chain-of-custody thinking, applied to AI.

  • Engineers, not slideware

    The people who scope the work are the people who write it.

  • Colorado-based, Colorado-law-aware

    We are in Conifer and work across the state, remote or on site. We help you get ready for Colorado's ADMT law.

  • A clear engagement model

    Assessment, a pilot in weeks, build, training, support. A fixed scope for every stage.

Colorado SB 26-189 · Automated Decision-Making Technology

Jan 1, 2027

The day Colorado's ADMT duties start

Signed May 14, 2026, SB 26-189 repeals and re-enacts the 2024 Colorado AI Act (SB 24-205). It covers AI used to materially influence consequential decisions about education, employment, housing, financial or lending services, insurance, health-care services, or essential government services. It is enforced by the Attorney General.

  • Developers give deployers documentation.
  • Deployers give consumers notice, explain the AI's role after an adverse outcome within 30 days, and honor data-correction and human-review requests.
  • Both keep records for at least 3 years.

We help you get ready: an inventory of where AI influences consequential decisions, consumer notices, human review and records.

EBOXLAB is not a law firm and this is not legal advice.

[ who's behind it ]

A forensics engineerwho builds AI.

Deniss Gladkihs

"Dan G" · founder, EBOXLAB

Evidence work teaches one habit above all: be able to show where every answer came from. That habit is in every AI system we build.

Dan G on LinkedIn
forensics
25 years in digital forensics and data recovery: failed, damaged, encrypted and deleted media, and court-ready evidence work.
software
Software architect behind a suite of AI platforms for legal professionals.
book
Author of Chain of Truth: Digital Evidence and the Future of Proof, available on Amazon.
education
Degree from an aviation university.
company
EBOXLAB, a Colorado IT, security, AI and digital-forensics company, in business since 2004.

[ faq ]

Nine questionswe hear first.

q01

Can AI run without sending our data to the cloud?

Yes. We deploy models on your own servers, including networks with no internet connection.

q02

Do we need a lot of data to start?

No. Document search works with the files you already have.

q03

How long does an AI project take?

A focused pilot typically runs a few weeks. Larger builds are scoped after it.

q04

What about wrong answers?

We design for them: answers cite their sources, staff can flag errors, and sensitive actions keep a person in the loop.

q05

What can you automate?

Processes that run on documents and repeated steps: intake, extraction, validation, routing and reporting. We build them as AI workflows and pipelines, connect them to the systems you already use, and keep a person approving where it matters.

q06

Who owns what you build?

You do. The code, models, prompts, data and documentation stay with you, in your own accounts. There is no lock-in.

q07

Which AI tools do you train teams on?

The ones your team already uses: ChatGPT, Microsoft Copilot, Claude and Gemini, plus the tools we build for you. Sessions are hands-on and run on your team's own workflows, with a leadership track and a staff track.

q08

Does Colorado's AI law apply to us?

It can, if AI materially influences consequential decisions about education, employment, housing, financial or lending services, insurance, health-care services, or essential government services. Duties under SB 26-189 start January 1, 2027. We help you get ready: an inventory of where AI influences those decisions, consumer notices, human review and records. EBOXLAB is not a law firm and this is not legal advice.

q09

Where are you, and do you work on site?

EBOXLAB is in Conifer, Colorado, and has been in business since 2004. We work with companies across Colorado, remote or on site.

[ contact ]

Book a call.Tell us what the AI has to do and what data it would touch.

emailinfo@eboxlab.com call · Colorado(720) 222-0110 call · toll-free(888) 714-6292

Conifer, Colorado · Denver metro and the Front Range · companies across Colorado, remote or on site