What problem needs solving, for whom and why?
Down-to-earth AI.
From Biert, in the Pyrenees, we build software and AI agents for real work. A simpler task, a tool that holds up, a team that knows how to use it.
What you receive.
What you decide.
A shared problem, a first version you can try, evidence to decide on.
The three documents
What do we deliver and how will we judge it?
Does the result work on this version?
Brief
- In this document
- Current situation, users, expected result, constraints, unknowns.
- Your approval
- Your business contact confirms the problem and the priorities.
Specification
- In this document
- Journeys, scope, rules, deliverables, acceptance criteria, usage budget, responsibilities.
- Your approval
- One identified version is the reference for the next batch.
Acceptance and delivery decision
- In this document
- Tests, results, evidence, defects, reservations, corrections and decision.
- Your approval
- The result is accepted, accepted with reservations or sent back. Publishing still needs its usual authorisation.
We work in small batches. You see and approve each result.
- Understand your task.
- Show a first version and try it together.
- Adjust, approve and go live.
- The three documents, up to date.
- An identified version of the tool that you can try.
- The access, the operating instructions and the way to roll back.
You set the priorities and approve the scope. You then decide whether the result can be delivered and published.
We try the agreed version and show you the evidence. You accept the result, ask for corrections or approve with reservations.
Publishing waits for your agreement.
A first useful result.
Choose the offer closest to your task.
Tasks handled by agents, under your control.
A business process takes up your time and depends on manual follow-ups.
- You receive
- An action ledger. Exceptions sent for human approval. Going live, and training for the team.
- Support
- A project package, then a subscription covering maintenance, hosting and API costs within the agreed scope.
You can cancel the subscription with one month’s notice. You keep the action ledger and your data.
Rank your uses of AI and test a first case.
You see possibilities with AI but do not know which to pursue.
- You receive
- A prototype to test the chosen case.
- Support
- A one-off engagement, with no subscription.
The engagement ends when we hand over the ranked use cases and the prototype.
A plan and exercises for using AI as a team.
Your team wants to understand AI and use it under the right conditions.
- You receive
- A session to inspire, exercises on your own tasks and reusable tools.
- Support
- One-off sessions, with no subscription.
The support ends after the session. The exercises and tools stay with the team.
Your voice on LinkedIn, with drafts for you to approve.
You have ideas to share on LinkedIn, but writing comes after everything else.
- You receive
- LinkedIn drafts for you to review.
- Support
- A monthly subscription.
You can cancel the subscription each month.
The right tenders, and prepared responses.
Finding the right tenders and preparing the files takes up your team’s time.
- You receive
- Prepared response files, with the supporting documents to check.
- Support
- A setup, then a monitoring subscription.
You can cancel the subscription with one month’s notice. The prepared files stay with you.
Your uses of AI tracked in an up-to-date register.
AI uses are multiplying and you lack a shared overview.
- You receive
- An inventory of systems and uses, a charter and tracking of the obligations that apply.
- Support
- A project package, then a subscription covering maintenance and upkeep of the register.
You can cancel the subscription with one month’s notice. The register is handed over in an open format.
A website or shared space adapted to your work.
Your activity needs a shared space that your current tools do not provide.
- You receive
- Depending on scope: accounts, applications, spaces and editable content. Launch and onboarding.
- Support
- A project package, then a subscription covering maintenance and hosting.
You can cancel the subscription with one month’s notice. The code and the data are handed over to you.
Paths to explore
Test an idea on a real case before you invest.
You want to see an idea working before you start a project.
- What we explore
- A demo on a journey you choose: try the interaction, see the limits and check what works.
- To scope together
- The scenario to show, the data available and the criteria for judging the trial.
The results help you decide whether a project is worth pursuing.
Find, compare and extract useful information from your documents.
Your team spends time rereading files to find or compare a piece of information.
- What we explore
- A trial on your documents: extract data, compare versions or search for an answer and trace it to its source.
- To scope together
- The formats, the quality of the scans, confidentiality and the checks you keep in your own hands.
The trial lets you judge reliability on your own documents before going further.
Explore AI together, starting from one of your team’s tasks.
Your team has ideas for using AI and wants to test them against its daily work.
- What we explore
- A workshop around a real case: try the tools, compare the results and choose a useful first trial.
- To scope together
- The starting task, the participants and a working format that suits the team.
You leave with the leads you chose and the points to check in the next trial.
Estimate the effect of giving a task to an agent
Change the assumptions: the trail recalculates as you type.
Result per week
- Time freed
- 8h
- Review time
- 2h
- Usage cost
- €10.00
- Net effect
- €350.00
Time handled by the agent: 10h per week.
How is the estimate calculated?
An estimate based on your assumptions, to check against your real task.
Time freed = time handled − review timeNet effect = time freed × hourly cost − usage costPer week: time freed 8h, review time 2h, usage cost €10.00, net effect €350.00. On these assumptions, the project pays for itself.
Tell us what you need
Is there an idea to build, or a task eating your days? Tell me what you would like to change.
Cairn
- You answer in your own words.No form: a conversation, at the pace of your task.
- Cairn notes what it understood.Facts, need, unknowns and next step, in plain sight.
- You review before anything is sent.Nothing goes to the Cairn team without your agreement.