Base
2 years
One review at a time, for two whole years.
Regularly $29/month
$10/month
Pay $240 once · 24 months
- 1 review at a time
- τ² AI reads up to 300 papers a month
- You and 1 co-lead
- Every step, import to write-up
Tau Square Studio · Only for our students
Your whole meta-analysis.
Your whole meta-analysis.
One workspace.
Studio takes you from search results to a submitted paper in one place, with τ² AI reading alongside you and guardrails at every step. So you do more meta-analyses, faster, without the mistakes that sink papers.
Learn the method in 100 hours with the course. Then run review after review in Studio.
The usual way
One review. Six tools.
One review. Six tools.
A hundred places to slip.
So here's how most students run a meta-analysis today: screening in one app, extraction in a spreadsheet they built themselves, risk of bias on paper forms, the analysis in a program they're still learning, the PRISMA diagram drawn on a slide, and a co-reviewer's work arriving by email. Every hand-off is a chance for a number to change on its way. And that one wrong number is the one a peer reviewer finds.
Before: six tools, hand-offs everywhere
Numbers copied by hand between them, again and again.
With Studio: one workspace
- Import and remove duplicates
- Screen, blind, with a partner
- Full texts, fetched and matched
- The coding sheet, locked
- Extraction, with sources
- Risk of bias and certainty
- The analysis, in real R
- The write-up pack · ready to submit
Every number travels by itself: from the paper, to your sheet, to your forest plot.
One workflow
Every step of the review,
Every step of the review,
in the order you learned it.
Studio follows the 100-Hour Protocol step by step. Each step hands its work to the next, so nothing is retyped and nothing gets lost.
-
Step 1
Import, and duplicates gone
Bring in your results from PubMed, Embase, Web of Science or any database that exports RIS or CSV. Duplicates are found and set aside, never deleted.
Kills: the afternoon lost to cleaning a messy library.
-
Step 2
Screening, blind and fast
Keyboard shortcuts, a calibration round before the real thing, and agreement (kappa) measured for you. The PRISMA flow draws itself as you go.
Kills: screening that drags on for weeks.
-
Step 3
Full texts, found for you
Free full texts are fetched for you. Drop the rest into a folder and Studio matches each PDF to its study. Retracted papers are flagged, and one trial published twice is caught.
Kills: the PDF chase, and counting one trial twice.
-
Step 4
The coding sheet, built and locked
Your extraction sheet is built from your protocol and the outcomes your papers actually report, the Divergent Data Extraction Protocol way. Approved, versioned and locked before extraction starts.
Kills: changing your sheet halfway and redoing everything.
-
Step 5
Extraction: verify, don't type
τ² AI points to every value in the paper, our algorithm reads and checks it, and you verify it with a keypress instead of typing it. Two extractors, compared and reconciled.
Kills: typing hundreds of numbers by hand, and the typos that come with them.
-
Step 6
Risk of bias and certainty
RoB 2, ROBINS-I, Newcastle–Ottawa, QUADAS-2 and AMSTAR 2, with each tool's own rules applied for you. Traffic-light plots, and a GRADE summary of findings.
Kills: bias tables that don't match your own judgements.
-
Step 7
The analysis, in real R
Point and click, on R and metafor, the package behind thousands of published meta-analyses. Forest and funnel plots, subgroups, meta-regression, sensitivity analyses, and the R script that reproduces every number.
Kills: the fear of code, and results you can't reproduce.
-
Step 8
The write-up pack
Your Methods and Results drafted from what you actually did, with references. The PRISMA 2020 diagram, the summary-of-findings table, an AI-use statement and a full audit trail.
Kills: writing your methods from memory, weeks later.
Meet Tau, the face of τ² AI.
τ² AI · The Glass Box Protocol™The AI points.
You decide.
τ² AI is the reading assistant built into Studio, and it never types a single value into your data. It works by our Glass Box Protocol™: τ² AI points to the answer in the paper, our own algorithm reads it and checks it, and you verify it. Nothing is hidden in a black box, so every value can be traced back to its page. That's what makes it defensible.
- 1τ² AIPoints.It finds where the answer is: the exact quote, the page, the table cell. That's all it does. It never types a value into your sheet.
- 2Our algorithmReads and checks.Plain code, with no AI in it, reads the paper itself: the quote must really be on that page, the number must really be in the quote, and the table cell must say what τ² AI says. Anything that doesn't match is flagged red.
- 3YouVerify.Studio highlights the source in the paper and you confirm it: one keypress when everything matches. Nothing enters your data until you do.
No hallucinations in your data. A number that isn't in the paper can't slip through: our algorithm catches it and flags it red, before you ever confirm a thing.
Less to hold in your head
You check one highlighted sentence, not a 30-page paper, and the colour tells you how hard to look: green is one keypress, red means look closely. Your attention goes where it's needed, so you stay sharp to the last paper.
Less blind trust in the machine
Automation bias is the pull to accept a computer's answer without checking it. The Glass Box Protocol™ is built against it: you always see the evidence, not just the answer; anything doubtful is flagged red, so it can't be waved through; and on eligibility, τ² AI shows its view only after you've made your own call.
Where τ² AI saves you hours
- It checks each full text against your eligibility criteria, and shows you its view only after you've decided, so it never sways you.
- It maps every outcome reported in every paper, so your coding sheet starts complete.
- It points to every value your extraction sheet needs, cell by cell, for our algorithm to read and you to verify.
- It finds the facts behind risk of bias: how groups were randomised and concealed, the losses in each group, and whether the trial was registered before it began.
It never calculates, never judges risk of bias and never writes your results. When a peer reviewer asks where a number came from, you show them the page. And because journals ask you to disclose AI use, Studio writes your AI-use statement for you.
Guardrails
Guardrails that catch
Guardrails that catch
what reviewers catch.
Studio won't let the classic mistakes through. When something's wrong, it stops you and tells you why, and if you have a good reason to go ahead, your reason goes on the record.
- Blind screeningNeither of you sees the other's decisions until you've both finished.
- Calibration firstA practice round on the same records, before the real screening starts.
- Agreement, measuredKappa on blind decisions only. If extraction agreement drops below 0.8, the Red Light protocol starts.
- Nothing unverifiedNo AI value reaches your analysis until a person has checked it.
- A locked sheetExtraction starts only after the coding sheet is approved and locked.
- Safe poolingNo pooling with fewer than two studies, and no funnel plot with fewer than ten.
- Number traps caughtA standard error typed as a standard deviation, a shared control group counted twice, randomised and non-randomised studies pooled together.
- A full audit trailEvery decision, who made it and when, kept for good.
The course and Studio
Learn it in 100 hours.
Learn it in 100 hours.
Then keep going.
The course teaches you the method, step by step, on your first meta-analysis. Studio is that same method built into software: the same steps, in the same order, with the same rules as the playbooks. So the next one doesn't start from zero. You already know what every step means, and Studio does the fetching, the typing, the formatting and the drawing. You do the thinking.
Your first reviewWhile you learn · about 100 h
Your secondWith Studio
Your thirdWith Studio
And the nextWith Studio
The playbooks, built in
- The Divergent Data Extraction Protocol becomes your coding sheet, with its tabs, its naming and its golden rule: missing means blank.
- The GIGO Protocol becomes your risk-of-bias and certainty steps.
- R-MetaScript becomes the R script you download with your analysis.
Your team
Bring your co-lead.
Every Studio plan comes with one co-lead: the research partner who screens and extracts alongside you. Your co-lead works inside your reviews with you, τ² AI included, and every two-reviewer rule in the protocol is covered by the two of you.
Your co-lead can't start reviews of their own. That's what their own Studio plan is for.
Students only
Studio isn't sold to the public.
Studio isn't sold to the public.
Not to anyone.
It's ours, it's built on our method, and it's only for Tau Square students. If you want Studio, you join the academy. Simple as that.
Studio plans
Add Studio when you join.
Add Studio when you join.
It's the lowest price it will ever be.
Studio is sold separately from the course. Add it in the same checkout when you join, and you get it for far less than the regular price. You pay once, upfront, with no auto-renewal.
Base
1 year
One review at a time.
Regularly $29/month
$20.75/month
Pay $249 once · 12 months
- 1 review at a time
- τ² AI reads up to 300 papers a month
- You and 1 co-lead
- Every step, import to write-up
Plus
1 year
Five reviews on the go at once.
Regularly $59/month
$41.58/month
Pay $499 once · 12 months
- 5 reviews at a time
- τ² AI reads up to 800 papers a month
- You and 1 co-lead
- Every step, import to write-up
Pro
1 year
As many reviews as you can run.
Regularly $100/month
$70.75/month
Pay $849 once · 12 months
- Unlimited reviews
- τ² AI reads up to 1,500 papers a month
- You and 1 co-lead
- Every step, import to write-up
Adding Studio later, after you've joined? The regular price is $29 a month for Base, $59 for Plus and $100 for Pro.
Choose your course plan, then pick your Studio plan in the same checkout: one payment.
- No auto-renewal: you pay once
- Always the latest version of Studio
- A GST invoice for your research or academic funds
Prices in US dollars. In India, 18% GST is added and shown before you pay. Studio runs in Chrome or Edge on a laptop or desktop.
Questions
Studio, answered.
Is Studio included in the course?
No. Studio is sold separately, as an add-on for Tau Square students. The best time to add it is at checkout when you join: that's the lowest price it will ever be. You can also add it later, at the regular price.
Can I buy Studio without joining the academy?
No. Studio is only for Tau Square students. It's built on our method, and we keep it that way.
What is τ² AI, and can I use it for published research?
τ² AI (Tau Square AI) is the reading assistant built into Studio, and it never types a value into your data. It works by our Glass Box Protocol™: τ² AI points to the answer in the paper, our own algorithm (plain code, no AI) reads it and checks it against the paper, and you verify it. A number that isn't in the paper is flagged red, so hallucinations don't get into your data. Journals ask authors to disclose AI use, so Studio writes your AI-use statement for you.
Where are my reviews kept?
In a folder you choose on your own computer, not in our cloud, so your work stays yours. Studio runs in Chrome or Edge on a laptop or desktop.
Who can work on my reviews with me?
You and one co-lead. Your co-lead screens and extracts alongside you, τ² AI included, but can't start reviews of their own.
How much reading does τ² AI do on my plan?
Up to 300 papers a month on Base, 800 on Plus and 1,500 on Pro. A paper counts once a month, however many times τ² AI reads it, and a paper longer than about 30 pages counts as more than one.
What happens when my Studio plan ends?
Your reviews stay in your folder, and you can still open and export them. Starting new reviews, τ² AI and working with your co-lead stop after a 14-day grace period.
Your next step
Studio is for Tau Square students.
Studio is for Tau Square students.
Become one.
Start with Phase 0. It's free: 14 short lessons, about 2 hours, no card needed. When you join, add Studio in the same checkout.