Assess & plan
Work out what is worth building.
For a team deciding where AI fits, what it will cost, or whether existing tools can do the job.
Deliverable: a requirements brief, baseline assessment, options, and a recommended project scope.
Badlands Labs · Calgary, Canada
Evaluate AI models. Plan training. Choose where inference runs.
Start with the task and the tools you already use. We compare AI with workflow changes and existing software before recommending a build. Some projects stop at research or planning; implementation is scoped only after the data, evaluation, and operating plan are clear.
AI work, scoped to the organization and the task.
Research & model evaluationTraining research & inference planningOn-premises & managed feasibility
Ways to engage us
Start with the work you need done.
Each engagement is scoped and quoted individually. Your proposal sets out professional fees, third-party costs, payment milestones, and any ongoing support.
Implementation planning
Move from a tested fit to a defined project.
For an organization with an agreed requirements brief and a clear owner for the system.
Deliverable: a bounded implementation scope, acceptance criteria, operating responsibilities, and handover plan.
Adoption support
Help a team use an agreed system well.
For staff who need guidance on the selected workflow, data handling, and output review within a scoped AI project.
Deliverable: user guidance, workflow handover, and defined follow-up.
Services
Specialist work, with a clear deliverable.
Engage us for a defined piece of work or bring us into an existing project. We agree on the responsibilities, deliverables, and acceptance criteria before implementation begins.
On-premises inference feasibility
For workloads with a clear reason to keep inference on-site.
A local model is only one part of the system. We first assess the workload, model and runtime, target hardware, data flows, user access, integrations, and operating owner. A deployment scope is considered only when quality, performance, maintenance, and support can be checked against the requirements.
Examples to consider during feasibility work include AMD Ryzen AI workstations, NVIDIA DGX Spark, and ASUS Ascent GX10. Suitability depends on the workload and current product specifications.
What the work includes
- Workload and hardware sizing
- Data-flow and integration design
- Local inference evaluation on agreed tasks
- Deployment, maintenance, and handover scope
Company knowledge systems
For teams spending too much time finding information.
We assess whether an internal wiki or searchable knowledge base fits the job, then plan content cleanup, access rules, source links, and updates. Document quality and a clear owner for keeping it current matter as much as the search tool.
What the assessment covers
- Document quality and ownership
- Wiki, search, and retrieval options
- Source links and access rules
- Who maintains the material and how
AI workflow review & adoption support
For teams unsure which models and workflows justify their cost.
Compare cost per completed task and output quality across your current setup and relevant alternatives. Review model choice, repeated context, and workflow steps, then guide the relevant staff on practical use, data handling, and output review as part of the agreed work. Any savings depend on the workload and the changes that prove useful.
What the work includes
- Usage audit and cost per completed task
- Model selection, caching, and workflow improvements
- Guidance for staff using the agreed workflow
- Standards for tool use, data handling, and output review
Model training research & edge inference feasibility
For specialized tasks and devices with limited resources.
Training is a later option, not the starting point. We compare an existing model and workflow changes against a real task. If a measured gap remains, we can scope a paid adaptation study using authorized examples and an agreed test. It may show that fine-tuning does not help.
What the work includes
- Task and data/model-rights review
- Existing-model and workflow baseline
- Adaptation options and independent evaluation plan
- Serving and support feasibility
How we deliver
Based in Calgary.
Scoped before work begins.
Badlands Labs is based in Calgary and takes on individually scoped AI projects for Canadian organizations. Work can include model evaluation and research into task-specific training, feasibility planning for customer-controlled or managed inference, and internal knowledge systems. The agreed scope identifies the delivery lead, roles, deliverables, and responsibilities.
We agree on access, data handling, ownership, and procurement requirements before work starts. The handover documents the system and the operating responsibilities your team accepts.
- Model research & evaluation
- Compare candidate models against representative examples and agreed task requirements.
- Training research & inference planning
- Evaluate whether adaptation is warranted for a defined task, then establish what serving and maintenance would require.
- On-premises & managed feasibility
- Assess deployment and support options around the organization's workload, controls, and capacity.
How we work
A defined process from brief to handover.
You know what is being delivered, who is responsible, and what needs approval at each stage.
01 / Enquiry & discovery
Understand the assignment.
We review your brief, confirm fit and availability, and arrange a scoping conversation. We identify stakeholders, constraints, and the outcome you need.
Record: project brief and open questions.
02 / Scope & agreement
Agree on the work.
The statement of work sets deliverables, exclusions, milestones, fees, client responsibilities, and acceptance criteria. Access starts after the required agreements are in place.
Approval: agreed scope and project owner.
03 / Delivery & review
Build against the brief.
We work in agreed milestones, share progress and issues, and demonstrate results against representative tasks. Changes to scope, fees, or timing require written agreement.
Review: demonstrations, test results, and decisions.
04 / Acceptance & handover
Leave a usable system.
Your project owner reviews the agreed criteria. We record outstanding items, transfer the deliverables, train the relevant staff, and confirm ongoing support arrangements.
Handover: runbook, training, and acceptance record.
A named lead on each sideYour project owner and our lead agree on the reporting cadence, decision process, and escalation contact at kickoff.
Access agreed before deliveryWe establish approved systems, data sources, and confidentiality requirements before accessing company material.
Documented scope changesNew requests are assessed for cost, timing, and dependencies before they become part of the assignment.
A planned exit and support arrangementOwnership, third-party licences, credentials, maintenance, and contractor access removal are covered in the scope and handover.
Before we start
Common questions.
Do we need to buy new hardware?
We assess your existing equipment first. If new hardware is needed, we size it for the models, number of users, and response times you require. The proposal separates equipment costs from setup and support.
Will our data leave our premises?
On-site data locality depends on the model, runtime, integrations, updates, and support needs. We map where inputs and outputs flow in the proposed design and identify any external services.
When is model training worth exploring?
We consider training only when existing models and workflow changes miss an agreed requirement. If data rights and representative examples are in place, we can test an adaptation against a baseline. It may not improve the result.
How do you assess potential savings?
We measure the cost of representative tasks with your current setup, then test alternatives against the same quality requirements. We include hardware, energy, maintenance, and setup costs. Whether an alternative saves money depends on the workload and deployment.
How are projects priced and approved?
We scope the assignment before quoting. A defined implementation can use a fixed project fee; exploratory or advisory work can use an agreed time budget. The proposal states the fee basis, payment milestones, expenses, and exclusions. Work starts after the scope and commercial terms are agreed.
Project enquiries
Bring us the assignment.
Tell us what needs to change, the systems involved, and your target timing. A short brief is enough to start a scoping conversation.
info@badlandslabs.com
Prefer to email directly? Include your organization, the intended outcome, and any delivery constraints. We will review the fit and discuss the next step.
Badlands Labs
Calgary, Alberta, Canada