Built from real research work
Research you can explain and defend.
Local-first tools to check claims, test assumptions, and document the decisions behind your work.
Start with the work in front of you
-
Plan
-
Analyse
-
Check & write
-
Review
Made inside real research work
I’m Heidi Andersén, MD, PhD, a clinician-scientist in Finland. Vahtian began with recurring problems in my own work: decisions that vanish, claims that drift from their sources, and methods that become difficult to explain later.
Practice comes first. I build each tool around work a researcher must actually do, then keep the steps visible enough to question, repeat, and improve.
A simple working rule
You decide
AI can advise, but it does not silently become the research decision.
Keep the record
Inputs, outputs, dates, and human choices stay visible enough to inspect later.
Keep data local
Your manuscripts, interviews, and working files stay on your machine whenever the workflow allows.
State the limits
The tools check support and consistency. They do not decide scientific truth.
Research kits
The free tools let you test the approach. The kits give you the templates, local workflow, and records to run it on a real study.
Complete workflow · qualitative studies
QualiVahti Local
A local qualitative workflow for sensitive interviews: transcribe, code, review suggestions, maintain the codebook, and export an audit trail without sending the data away.
Explore QualiVahti Local · €49 →Complete workflow · observational studies
StudyVahti Vault
Keep the research question, variable dictionary, estimand, analysis plan, decisions, outputs, manuscript claims, and 16-check audit trail in one local workspace.
Explore StudyVahti Vault · €49 →Focused kit · qualitative sampling
MethodVahti Sample Defence Pack
Set the stopping logic, keep an adequacy log, and draft a defensible sampling paragraph without treating saturation as a formula.
See the Sample Defence Pack · €49 →What I’m proud of
Open-source research packages with methods engines in Python and R. The code is there to inspect, test, and reuse.
AssessLite open source · R + Python
Stress-test the assumptions behind a causal result across clusters, time, subgroups, confounding, DAG adjustment, and overlap. Export an audit record of what held, failed, or could not be resolved. Apache-2.0.
Where the tools stop
You make the research decision
Vahtian helps you inspect evidence, organise work, run checks, and keep a record. It can point to a problem. It cannot decide whether a claim is true, a study is sound, or a result should be trusted.
See what happened
Checks, calculations, and exports leave a record you can read and run again.
Treat AI as a second reader
AI can flag a concern or suggest a direction. You decide what counts as evidence and what enters the research record.
Keep judgment with a person
Decisions about inclusion, interpretation, and reporting stay with the researcher or reviewer responsible for them.
Run it locally. Inspect the code.
The browser tools and open-source packages run on your laptop where possible. Most use Apache 2.0; each repository states its licence. Checking your own work will not sit behind a paywall.
vahti, Finnish for sentinel / guard.
Need Vahtian for your team?
I work with research groups that need a shared workflow, a clear record, or support putting the tools into practice.
-
Supervisors and doctoral schools
Team licences for QualiVahti Local and StudyVahti Vault.
-
Guideline teams
-
Journals and research organisations
Bring a workflow, integration, or research partnership question.