How we work
Clear scope, human oversight, measurable outcomes.
- Engagement
- Fixed scope
- Payment
- 50 / 30 / 20
- Approval
- Human
- Handover
- Documented
The engagement loop
Discover → Design → Build → Improve, then again
- DiscoverUnderstand the process before proposing a tool. You receive: A written problem statement and a shortlist of options, including the option of doing nothing.
- DesignChoose the smallest system that solves the problem. You receive: A signed scope with deliverables, exclusions and a fixed payment schedule.
- BuildShip in visible increments, with your team involved. You receive: A working system in your accounts, with documentation and a handover session.
- ImproveMeasure the outcome, then decide what is next. You receive: An outcome report and a decision on whether to move to a managed service.
- Discover: Understand the process before proposing a tool. You receive: A written problem statement and a shortlist of options, including the option of doing nothing.
- Design: Choose the smallest system that solves the problem. You receive: A signed scope with deliverables, exclusions and a fixed payment schedule.
- Build: Ship in visible increments, with your team involved. You receive: A working system in your accounts, with documentation and a handover session.
- Improve: Measure the outcome, then decide what is next. You receive: An outcome report and a decision on whether to move to a managed service.
The process
Four stages, same shape every time.
Discover
Understand the process before proposing a tool.
We start with how the work is done today: the steps, the handoffs, the exceptions and the places where time disappears. No platform decision is made in this stage.
- Stakeholder interviews with the people doing the work
- Current-state process map with owners and handoffs
- Baseline measurement on the metric we intend to move
You receive
A written problem statement and a shortlist of options, including the option of doing nothing.
Design
Choose the smallest system that solves the problem.
We map the data, permissions and approval points before interface work. If an off-the-shelf tool fits, we say so — a custom build has to earn its place.
- Solution options with cost, risk and trade-offs
- Data model, access rules and approval gates
- Scope, acceptance criteria and milestone plan
You receive
A signed scope with deliverables, exclusions and a fixed payment schedule.
Build
Ship in visible increments, with your team involved.
Work happens in short milestones. You see the system before it is finished, and anything that requires a decision gets raised while it is still cheap to change.
- Implementation in milestones with review points
- Testing against the acceptance criteria, not just a demo
- Permissions, logging and error handling from the start
You receive
A working system in your accounts, with documentation and a handover session.
Improve
Measure the outcome, then decide what is next.
After launch we compare against the baseline, fix defects, and agree what should be optimised. Expansion only happens when the first step produced evidence.
- Outcome report against the original baseline
- Defect resolution and small adjustments
- A written recommendation: optimise, expand, or stop
You receive
An outcome report and a decision on whether to move to a managed service.
Engagement types
Four ways to start, with a fixed endpoint.
1 session
Paid discovery
A structured session and short written recommendation. Credited against the project if we proceed.
2–8 weeks
Fixed-scope project
Website and CRM, knowledge assistant or custom platform. Milestones and acceptance criteria agreed upfront.
2–4 weeks
AI workflow pilot
One workflow, measured against a baseline, with an approval queue before external actions.
Monthly · 3-month minimum
Managed operations
Monitoring, adjustments and support for what we delivered. Monthly, with stated usage allowances.
Commercial terms
What the paperwork says, in short.
- Discovery
- Paid discovery session, credited against the project fee if we proceed.
- Project payment
- 50% upfront, 30% at milestone, 20% on acceptance.
- Pilot scope
- Fixed scope and written acceptance criteria before work starts.
- Monthly service
- Billed monthly in advance, with usage allowances stated separately.
- Minimum term
- Three months for managed AI operations, where it applies.
- Change requests
- Estimated and approved before any additional work begins.
- IP ownership
- Source code, configuration, customer data and reusable components defined in the agreement.
- Data & security
- Data processing terms where required, least-privilege access, and retention rules agreed in writing.
These are commercial terms, not legal advice. For EU/UK or Australian clients, data processing terms and applicable obligations are reviewed before the agreement is signed.
Security & data
What is enforced in the build.
- Tenant isolation and authorisation enforced in the backend, not the interface
- Document-level permissions when a source must be restricted
- Source references on every answer grounded in a knowledge base
- Tool calls and agent actions logged with enough context to review
- Approval required before sending email, changing CRM records or taking external action
- Usage limits per client, with inference cost visible
- A documented process for deleting documents and revoking access at contract end
- Backup and restore tested before go-live
Principles
How we make decisions.
Service first, product second
We deliver the work before packaging it. Reusable systems come from repeated delivery, not from a roadmap written in advance.
Human approval before external action
Anything that leaves the business — an email, a proposal, a CRM change — passes a person first. Automation handles preparation, not judgement.
Evidence over promises
We take a baseline, measure after delivery and report what actually changed. No percentage savings quoted before the work is done.
Scope you can read
Every proposal lists what is included and what is not. Surprises in delivery are a design failure, not a negotiation tactic.
Questions
The practical questions.
What kind of firm is Samlution built for?
Do you sell software or services?
How does pricing work?
How quickly can we start?
What happens to our data?
Will AI act on our systems without approval?
What if AI produces something wrong?
Do you work with firms outside your two focus industries?
Next step
If the scope is clear, we can start.
The first call exists to find out whether it is. Thirty minutes, no slides.
No obligationFixed scope, fixed feeApproval stays human
See how engagements run