TANTEI

AI for the research process. Held to research standards.

Tantei helps research labs and institutes speed up the adoption and integration of artificial intelligence.

Talk to us
The OntologistONTOLOGISTOrganizes all the data from your studies, systematically.
The ProposerPROPOSERSuggests what is worth studying, out of what has already been collected.
The CriticCRITICFinds problems in a questionnaire before it goes into the field.
Coding questions agentRESPONDERHelps manage the tasks of research assistantsNext version
Codebook agentDOCUMENTERWrites the study's context document, from your instructions only.Next version
Team tracking agentOBSERVERWatches the team's sheet and reports, without writing to it.
Footnotes agentCITATORFinds relevant sources for citations, and pulls exact quotes from them.
Source collection agentFINDERBrings the material from the databases into one folder.
Document coding agentCODERTurns documents into a data table, with a quote for every value.
Grants agentSUBMITTERHelps submit to funders and journals
Citation styleFOOTNOTESFormats citations to the Israeli Uniform Citation Rules, in every format.
Privacy checkerPRIVACYCounts identifying details in the material before anyone touches it.
Lead agentCONDUCTORTakes a request in plain words and passes it to whoever needs it.
Survey coding agentSURVEYHelps run a survey end to end: advice, wording, fielding and collection.
Statistical analysis agentEXAMINERChooses and runs the right statistical test.
Style cardSTYLEAnalyzes how you write, and helps you draft papers and texts in the same style, easily.
Example
A · The ontology

What you already know. And what you have not tested yet.

Studies that succeeded or failed sit in folders on a computer. Tantei builds the index out of them: one ledger with a record for everything that was done, every document collected, every analysis run, and every study carried out.

  • TestedWhat has already been tested, and where the output that proves it is.
  • AskedWhat was asked, and what answer came back.
  • AssumedWhat was assumed along the way, without being tested.
  • Still openWhat was opened and never closed, and is waiting.
Example: a lab's scattered files become one index, a record per file with what is known about it, and out of the open records comes a proposal.
Example
Your folders
  • experiment.xlsx
  • interviews/
  • survey.docx
  • methods.docx
  • coding.xlsx
  • paper.pdf
  • draft.docx
  • plan.docx
The index
  1. Experiment dataTestedexperiment.xlsx
  2. InterviewsAskedinterviews/
  3. QuestionnaireAskedsurvey.docx
  4. Assumption: the sample is representativeAssumedmethods.docx
  5. Interview codingTestedcoding.xlsx
  6. Published paperTestedpaper.pdf
  7. Unpublished draftOpendraft.docx
  8. Analysis planned, never runOpenplan.docx
The proposal
Run the planned analysis

On data already collected. Nothing collected again.

Lenses pass over the index looking for one thing: what already sits in the data and was not the question it was collected for. Every such proposal comes with the records it grew from, and with one check you can run.

B · The agents

An AI agent platform for research processes

A range of agents for different roles. Each agent does one job, one agent conducts them all and connects them, and every result comes with the visual evidence behind it. Four of them, on demo data.

  1. Source collection agent

    You write what you are looking for, it suggests which databases are worth collecting from, and once you approve it collects everything into one organized folder, with a list of what was found and what was not.

    The source collection agent's screen: a wall of databases in five families, with a logo and a name for eachDEMO

  2. Document coding agent

    Turns a pile of documents into a data table. Every value the agent coded comes with visual proof.

    The document coding agent's screen: a document becoming a row in a table, with a quote under every cellDEMO

  3. Statistical analysis agent

    Chooses and runs the right statistical test, and shows each test's assumption as measured on your data.

    The statistical analysis agent's screen: bars and a journal-style results tableDEMO

  4. Survey coding agent

    Helps you run surveys end to end: from advice and pulling up surveys already done on the topic, through the wording, to fielding the survey and collecting the data, quickly and easily.

    The survey coding agent's screen: four stations on one rail, advice, what has been done, wording, fielding and collectionDEMO
DEMO
How it reaches you
  1. We arrive with the agent library
  2. We map your needs
  3. We install it with you.
C · Build

Is there something in your work you would like to be better? Give it to us.

We know how to build a range of AI-based solutions for research processes, and to fit them to your materials, your language and the way you work.

Requests we have already received
An agent that analyzes my social media ads
An agent that watches recordings and transcribes them
An agent that tracks protocols
An agent that classifies thousands of images

Tell us

D · Where the material goes

Full transparency in how the data is kept

Your computer Your files The tables The drafts Document coding agent Statistical analysis agent Privacy checker A person approves every result before it enters the table Only the text the model needs to read Model providerProcesses, does not keep The files do not leave
The files stay with you.
Human eyes at the right moments
An agent in charge of privacy
E · Talk to us

We want to help you work with AI, easily

Tell us in a few words what you research and what you would like solved, and we will get back to you for a conversation.

We will use these details only to get back to you. What happens to them.