The single-score model
?78%
- Unclear basis for the result
- Weak context for an appeal
- Pressure to treat probability as proof
Authorship evidence infrastructure
Evidence for authorship decisions that deserve scrutiny.
Viskopic is building a layered evidence and review system for higher education. We help institutions examine how a text was produced without reducing a consequential decision to one opaque score.
Pre-seed research and product development · Bristol, UK
For academic integrity teams, compliance leaders, and institutions preparing for an era of mixed human and machine authorship.
01 Institutional problem
A probability cannot carry the weight of a decision.
Academic integrity cases affect students, staff, and institutions. Yet many detection products collapse a complex question into a percentage that is difficult to interpret, challenge, or defend.
The Viskopic approach
Authorship contextTextual indicators
Reasoning reviewHuman decision record
A structured body of evidence a reviewer can inspect, question, and document.
02 Review workflow
From submitted text to a reviewable case.
Viskopic is designed as decision support for institutional teams. Analysis informs the process; accountable people remain responsible for the outcome.
-
01
Ingest with context
Collect the submitted text and the permitted comparison material needed for an institution's chosen review policy.
Submission record -
02
Build layered evidence
Examine authorship consistency, genericity, and the substance of the reasoning as distinct signals rather than one verdict.
Evidence profile -
03
Review through dialogue
Where closer scrutiny is warranted, use targeted questions to help a trained reviewer test understanding and provenance.
Structured viva -
04
Record the decision
Preserve the evidence considered, reviewer rationale, and outcome in a case record suitable for governance and appeal.
Auditable case
03 Evidence framework
Separate the questions before interpreting the answer.
Our research direction treats authorship as a set of related but distinct questions. Each layer should be interpretable on its own and assessed in context.
Core principle No metric should, by itself, determine misconduct.
Consistency
How does this text compare with the author's permitted prior writing?
Genericity
Where does the writing rely on unusually standardised or non-specific patterns?
Authenticity
Can the author account for the claims, sources, and reasoning in the work?
04 Research and governance
Designed for institutions that need more than a model output.
The hard problem is not only detecting patterns. It is building a process that can be evaluated scientifically, operated responsibly, and explained to the people affected by it.
Evaluation before assertion
Benchmarks should test performance across disciplines, writing conditions, model families, and student populations.
Human accountability
Software can organise evidence and generate questions. Institutions and trained reviewers own the final decision.
Proportionate data use
Collection and retention should be purpose-limited, policy-led, and designed around institutional control.
Contestable outcomes
Affected people should be able to understand what evidence mattered and how it informed the review.
Current programme
Building the evidence pipeline and institutional web application.
Data collection and storage Authorship analysis pipeline
Reviewer case workflow Pilot evaluation design
05 Company
Research, product, and institutional adoption under one thesis.
Viskopic is an early-stage company founded in Bristol. Our team combines applied AI research, data engineering, product development, and institutional partnership work.
SS
Visionary Founder
Sathvik Sai Krishna Surisetty
Leads the company thesis, institutional relationships, partnerships, and pathways to adoption.
AI Engineering, Lloyds Banking GroupBSc Data Science, University of Bristol
CS
Research Founder
Christopher Bintang Soo
Leads authorship research, model evaluation, benchmarks, and the scientific direction of the platform.
MSc Artificial IntelligenceBSc Data Science, University of Bristol
HN
Technical Founder
Hung Nguyen
Leads data pipelines, platform engineering, integrations, infrastructure, security, and reliability.
BSc Mathematics & Computer ScienceUniversity of Bristol
06 Development pathway
A focused path from research system to institutional infrastructure.
- Now
Evidence foundation
Data pipelines, metric evaluation, and the first end-to-end reviewer workflow.
- Next
Institutional pilots
Controlled deployments with partner universities and policy-led evaluation.
- Then
Integrated infrastructure
Secure integrations with learning, publishing, and document systems.
Start a conversation
Help shape a more defensible approach to authorship.
We are speaking with universities, academic integrity and compliance teams, research collaborators, and investors who share the need for evidence-led review.