AI agents
Systems that use your knowledge and tools to carry out a task, with defined access, limits, and review points.
Example
Research an account, gather the relevant context, and prepare a brief for your team.
Context · Tool use · Human review
AI transformation for the whole company
Build your
company brain.
We build AI agents and workflows and adapt models through post-training to fit your company’s knowledge, tools, and tasks.
Why the work gets stuck
Your company knows more than it can use. Let’s connect the pieces.
The answer is in a document. The context is in someone’s head. The next step is waiting in another tool. Your team spends time putting the pieces back together.
We work with you to connect that knowledge to the decisions and tasks it should inform. We call this connected system your company brain.
What we build
Built around
the job at hand.
Some tasks need an agent. Others need a reliable workflow or a model adapted to the task. We start with the work, then choose the approach.
Workflows
Connected steps that move work between people and systems, with clear rules for approvals and exceptions.
Example
Turn an incoming request into a draft response, route it for approval, then update the record.
Integrations · Handoffs · Approvals
Post-training
Adapt a model using task examples and feedback when evaluation shows a gap worth addressing through training.
Example
Fine-tune a model on reviewed examples, then test its output on cases it hasn’t seen.
Training data · Fine-tuning · Evaluation
The company brain
One intelligence.
Four capabilities.
Connected knowledge, AI agents, and workflows, with model post-training where it helps. Built around your systems, with people in control.
Your companybrain.
People in control01 /
Organizational memory
Bring documents, conversations, and decisions together so people can find answers with the source attached.
02 /
Decision intelligence
Put relevant history, rules, and tradeoffs in front of the person making the call.
03 /
Intelligent execution
Build agents that research, draft, and coordinate tasks, with workflows that route work to the right people for approval.
04 /
Continuous learning
Evaluate real tasks, learn from feedback, and use model post-training where it helps. Test improvements before rollout.
How we work
Start with one bottleneck.
Build from there.
01
Map the work
Together, we trace one important workflow: who does what, where context gets lost, and what a better outcome would look like.
02
Build a working system
We build the agents and workflows for that job, connect your knowledge and tools, and test them on real tasks with your team.
03
Make it part of the day
We refine the system with the people using it, document how it works, and agree what to improve or expand next.
The founders
Work directly with us.
We’re Logan H. and Matthew B. We’re building The Import Company to help teams turn their own knowledge into a more useful way of working.
LH
Logan H.
Co-founder
MB
Matthew B.
Co-founder
Before we talk
A few useful
starting points.
Do we need to replace our existing tools?
We start by looking at the tools and processes you already use. The goal is to connect useful context to the work, with any changes to your systems agreed as part of the scope.
Does every project need post-training?
No. We first define how we’ll evaluate the task and test a baseline. Better instructions, access to relevant information, or a clearer workflow may be enough. We consider post-training when the evidence supports it and suitable training data is available.
What should we bring to the first conversation?
One task that takes too long, loses context, or gets repeated often. Tell us who does it and which tools are involved. You don’t need a technical brief, and you don’t need to share confidential documents through this form.
Start a conversation
Where does work
get stuck?
Tell us about your team and one thing that takes more effort than it should.
We’ll use that context to explore a useful starting point with you. An inquiry starts a conversation; it doesn’t commit you to a project.