01
Verify before showing
Every reference is resolved against doi.org, Crossref, DataCite and OpenAlex before it reaches the page. One that can't be verified is removed, and the reply says so.
Stratide Labs
Research software that checks its own work.
We build tools for people whose work depends on getting the details right. Our first product is VERA, a research workspace in the browser. Every reference it writes is checked before it is shown, every statistic comes from the researcher's own data, and any choice that would change a result is left to the researcher.
VERA V1.1 is live and free during launch
01 · What we stand for
One rule, applied everywhere
An empty answer is better than a false one.
General AI writes references that were never published and describes numbers nobody computed. For a thesis or a grant application that is a professional liability. These commitments are how we keep the rule, and we enforce them in code rather than leaving them to a model's discretion.
02
Compute, never describe
Statistics are calculated over every stored row, with placeholder codes such as 9999 detected and excluded. Where nothing was computed, the answer says "not computed" instead of estimating.
03
The researcher decides
When a choice would change a finding, such as dropping duplicates or treating 9999 as missing, the software stops and asks. It sets out the options, what each one does and which it recommends.
04
Show the work
Each step of a request is shown as it runs, and a step that didn't run is never shown as done. Every analysis can name the exact version of the data it used.
05
Private by default
A researcher's rows never reach a language model; only the computed results are quoted. Accounts are kept separate, and everything can be exported or deleted.
06
Fair access
A free daily allowance for every researcher, and one free account per device, with a review path for shared lab and café computers.
02 · What we're building
One workspace for the whole of a research project
VERA Virtual Expert Research Assistant
Researchers usually carry a project between a word processor, a spreadsheet, a statistics package and a chat assistant. VERA puts the reading, data work, code and learning in one place, with one memory of the project, so it knows where the work stands before the first message of a session.
Researchers can type or speak to VERA and hear every reply read aloud.
Built for
- PhD and graduate researchers
- Lecturers and professors
- Independent researchers and think tanks
- STEM students
Chat
Research conversations with live web search, academic paper search and verified citations. Replies export to PDF, and their tables to Excel or CSV.
Document
PDF, Word and text files read in full and answered from retrieved passages, with page numbers. Read a paper to understand it, replicate it, find its gaps, critique it, extract from it or compare it with others.
Dataset
Real analysis on the researcher's own CSV and Excel files, with every sheet of a workbook loaded. A staged cleaning pipeline stops at each decision, and every dataset is kept as a version with a record of what changed.
Code
Plan first, then build module by module, with progress tracked against the plan. A build paused for a side conversation picks up at the next module.
Teach Me
A structured curriculum on any subject, taught subtopic by subtopic and assessed, with a report card, practice on weaker areas and reading found in real library catalogues.
03 · Mission
To give researchers software as careful as they are.
Research is sustained, multi-step work: reading, measuring, analysing, writing and learning. The tools around it were built for something else, and each hand-off between them costs time and lets an error in unnoticed. We build software that checks, computes and asks, so researchers can spend their time on the questions only they can answer.
“The long-term vision is not a better research assistant. It is an autonomous scientific intelligence.”
04 · Objectives
Where we're going
Our roadmap, in the order we intend to deliver it.
NowVERA V1.1, live
- Five workspaces, verified citations, staged data cleaning and research intents for documents.
- The VERA Early Access Group: 100 researchers shaping the next version.
- A free tier for researchers in the UK, USA, Canada, Australia, Germany, France and Nigeria.
NextVERA V2
- Paid plans for graduate students, independent researchers and heavy data users.
- A desktop app for PC.
- Deeper dataset tools and support for larger files.
LaterVERA V3
- Institutional deployments with seat management.
- Shared projects for research groups.
- Multi-step research tasks that VERA carries out end to end, with the researcher approving each decision.
05 · Contact
Tell us what VERA should become
VERA is free during launch. We'd like to hear from researchers, institutions, investors and collaborators who want to use it and help decide what comes next.