Industry playbooks
AI that knows what a manifest, an RFI, and a proxy packet actually are.
Generic AI advice is worth what you pay for it. Pick your industry and you get the specific workflows we build, the Alaska constraints they have to survive, and an honest list of what we will not automate. Fourteen playbooks, each written for the person who does the work.
Written by Kristian Baer, Northtek · Anchorage, Alaska · Updated 2026-08-21
The short answer
What does an Alaska AI consulting engagement actually produce?
A working automation inside a system your team already uses, plus the documentation to run it without us. We start by watching one real workflow end to end - dispatch, submittals, shareholder mail, prior authorizations, whatever is eating the most hours - then build the narrowest agent that removes the repetitive part of it and hands a person the judgment calls. Most first builds ship inside 30 days. Every output carries a citation back to the source record. You own the code and the data, and if you stop working with us the automation keeps running.
Fourteen playbooks
Find the one that sounds like your Monday morning.
Each page names six workflows, the artifact each one drops into your system, and the parts of your operation we would leave alone.
Logistics & Freight
Manifest intake, rate quoting, POD reconciliation, and detention billing for operators moving freight by road, barge, and air.
Rate quote drafting · BOL and manifest intake →
Construction & Contractors
RFIs, submittal logs, change order packaging, daily reports, and certified payroll for general and heavy civil contractors.
RFI drafting from the field · Submittal register management →
Alaska Native Corporations
Shareholder services, subsidiary reporting rollups, records search, and proposal support - on infrastructure the corporation controls.
Subsidiary reporting consolidation · Shareholder services triage →
Government Contracting
Solicitation triage, Section L and M compliance matrices, past performance libraries, and an honest CUI boundary.
Solicitation triage and go/no-go scoring · Compliance matrix generation →
Electrical & Mechanical
Service dispatch triage, T and M ticket capture, takeoff support, and prevailing-wage timecards for MEP contractors.
Service dispatch triage · T and M ticket capture →
Healthcare & Behavioral Health
Referral intake, prior authorization packets, recall outreach, and credentialing chase - with PHI handled under a signed BAA or not at all.
Referral intake completeness · Prior authorization assembly →
Oil, Gas & Mining
Permit and regulatory document search, HSE incident intake, procurement expediting, and turnaround readiness for remote operations.
Permit and commitment search · HSE incident intake →
Aviation & Air Carriers
Charter quoting, load manifests, maintenance document tracking, and dispatch communication for Part 135 operators.
Charter quote drafting · Load manifest preparation →
Seafood & Processing
Fish ticket reconciliation, crew settlements, seasonal onboarding, and regulatory reporting through a compressed season.
Landings and fish ticket reconciliation · Crew settlement preparation →
Utilities & Energy
Outage communication, work order intake, regulatory filing prep, and grant reporting for co-ops and rural utilities.
Outage communication · Field work order intake →
Tribal Governments & Nonprofits
Grant discovery and fit scoring, reporting assembly, council packet preparation, and member services for small administrative teams.
Grant discovery and fit scoring · Grant reporting assembly →
Professional Services
Client intake, document review with citations, billable time capture, and proposal drafting for law, accounting, and engineering firms.
Client intake and conflict preparation · Document review with citations →
Tourism & Hospitality
Inquiry triage, weather rebooking, seasonal staff onboarding, and multilingual pre-arrival for lodges, tours, and charters.
Inquiry triage and response · Weather disruption rebooking →
Real Estate & Property
Lead response, maintenance intake, lease abstraction, and owner reporting for brokerages and property managers.
Lead response and qualification · Tenant maintenance intake →
Where we work
On the road system and off it.
We are based in Anchorage and work statewide. We will fly out when being in the room matters, and we will not bill you for a flight when a screen share does the job.
What we built, in the open
One Alaska firm publishes the AI infrastructure it sells. This is it.
Every consultancy in this market says the same sentence about being technical. Only one of them has a compiler, a memory server, and a verification gate you can open in a browser right now and read.
-
Kryos
A programming language and compiler built so an agent's actions can be audited line by line. Full commit history, public toolchain, written here.
-
GENOME
An offline memory server benchmarked against a commercial competitor on public datasets, with the harness and the complete result set published.
-
FACTGATE
A verification gate for model output with its false-accept rate published, including the cases it does not catch.
Your build will be private and purpose-made. This exists so you can tell, before the first meeting, which kind of firm you are talking to.
Where your data goes
Four commitments that go in the agreement
Four commitments that go in every agreement, in every industry on this page.
- A named list of every system that will touch your data, and where each runs, before any code is written.
- Your documents are never used to train a model. Contract term, not policy page.
- Sensitive workloads run inside infrastructure you control - your tenant, your servers, or fully offline. We build the memory layer, so this is a configuration choice.
- You own the code, the configuration, and the data, with documentation written for your staff.
Straight answers
Do you actually come on site, or is this all remote?+
Both, and we are direct about which one your project needs. Mapping a dispatch floor or a shareholder services desk is worth being in the room for, so we come to you. Building, testing, and iterating afterward is faster over a screen share, and charging you for a seat on an Alaska Airlines flight to watch a deployment finish is not value, it is theater. We are in Anchorage, so getting to you is a short trip, not a project expense line.
How is this different from hiring a consultant who sets up ChatGPT for us?+
A tool rollout ends when the licenses are provisioned. What we build is a workflow that reads your documents, writes into your systems, and can be audited. We also build the layer underneath it - the memory server, the verification gate, the agent language - and we publish that code on GitHub so you can inspect the engineering before you trust it. That is the difference between configuring someone else’s product and being able to build whatever your operation actually needs.
What size company is this for?+
The math works when a task repeats more than roughly ten times a week and follows rules a person could write down. That describes a four-person dispatch office as easily as a Native corporation with a dozen subsidiaries. The scope flexes to the operation - a small company gets one workflow built tightly, a large one gets a program - and both get the same engineering underneath.
Who owns what you build?+
You do. Code, prompts, configuration, and data. We hand over the repository and the documentation, and we will train your IT staff to run it. There is no platform fee that holds your workflow hostage, and no clause that makes leaving expensive.
What if our data cannot leave Alaska, or cannot leave our building?+
Then it does not. We build offline memory infrastructure as part of our own research, so running a workload inside your tenant or fully air-gapped is a configuration choice rather than a special engineering project. For regulated data - CUI, PHI without a signed BAA, anything under a tribal data agreement - we scope what can run before we build, in writing.
How fast does something actually go live?+
A first workflow typically ships in 30 days: about a week mapping and instrumenting the real process, two weeks building and testing against your historical documents, and a week running it in parallel with your current process before anyone depends on it. We would rather ship one workflow that survives a busy week than five that need babysitting.
Bring one workflow. Leave with a build plan.
A working session is 60 minutes, free, and ends with a written recommendation: the workflow we would build first, what it recovers, and how fast it ships.
Anchorage, Alaska · info@northtek.io · (907) 903-4353