Step 1 of 6
Automated
Project
A workspace holds related analyses, cases, profiles, evidence, and results.
Proof by QuantBayes
Trust your result. Measure the evidence behind it.
QuantBayes Studio turns your analysis into evidence-based conclusions. A community agrees what evidence is needed to verify work. QuantBayes lets you fulfil that checklist and trust your results.
Step 2 of 6
Automated
Case
A case holds the items being evaluated, such as AI outputs, engineering checks, clinical findings, design decisions, claims, targets, or review tasks.
Step 3 of 6
User controlled
Evidence profile
A community or team defines the checklist of evidence required before scoring begins.
Step 4 of 6
User controlled
Evidence input
The user, pipeline, or agent adds evidence against each rule in the selected profile.
Step 5 of 6
Automated
Engine
QuantBayes scores evidence sufficiency and returns uncertainty for every item.
Step 6 of 6
Automated
Result
The result is saved with its profile, evidence matrix, and audit trail.
1
Project
Project
2
Case
Case
3
Profile
Evidence profile
4
Evidence input
Evidence input
5
Engine
Engine
6
Result
Result
Why it is trustable
A result is only useful when the evidence trail is visible.
1
Define what proof requires
A user, team, or community chooses the evidence rules that a result must satisfy.
2
Check the work against those rules
Each answer, claim, finding, or decision is reviewed step by step, including missing evidence.
3
Measure the confidence
The QuantBayes engine measures evidence sufficiency and saves the full trail for review, including confidence intervals.
Works with
AI outputs, agent workflows, clinical findings, engineering checks, research claims, design decisions, and regulatory review.
Evidence made visible
A simple story for a serious idea.
Use the scientific method as an operating system for decisions: define the evidence, test the result, update confidence, and keep the full trail. Apply the same formal method wherever evidence matters, from AI outputs and clinical findings to design decisions, engineering checks, and automated workflows.
1 of 6
Where it fits
One evidence layer for AI, science, engineering, and regulated work.
A profile defines what evidence is required. A case records the items being evaluated. A run saves the evidence matrix and the QuantBayes result.
AI answer verification
Check whether an answer is supported by declared sources, evidence rules, and inspection steps.
Agentic workflows
Require an agent to return evidence against each rule before a result is accepted.
Research review
Turn literature claims, methods, datasets, and assumptions into inspectable evidence checks.
Clinical and omics analysis
Review candidate findings against a versioned profile before ranking evidence sufficiency.
Engineering sign-off
Confirm that design, test, safety, and release criteria have been checked.
Regulatory evidence trails
Preserve the profile, evidence matrix, result, and report for later review.
API proof workflow
Add proof to any AI workflow.
QuantBayes Studio is not another model. It is an evidence layer that can sit beside any model, agent, pipeline, or human review process.
Declare the profile
The checklist is known before evidence sufficiency is scored.
Attach the evidence
Each item is checked against each rule using TRUE, FALSE, or NA.
Save the run
The output is stored with the profile, matrix, result, and audit trail.
API exampleDevelopers
import os
import requests
qb_base_url = os.getenv("QB_BASE_URL", "http://localhost:3000")
qb_api_key = os.getenv("QB_API_KEY")
response = requests.post(
f"{qb_base_url}/api/v1/runs",
headers={
"Authorization": f"Bearer {qb_api_key}",
"Content-Type": "application/json"
},
json={
"case_uid": "analysis_001",
"profile_uid": "ai_evidence_review_v1"
},
timeout=30
)
response.raise_for_status()
print(response.json())library(httr)
library(jsonlite)
qb_base_url <- Sys.getenv("QB_BASE_URL", "http://localhost:3000")
qb_api_key <- Sys.getenv("QB_API_KEY")
res <- POST(
paste0(qb_base_url, "/api/v1/runs"),
add_headers(
Authorization = paste("Bearer", qb_api_key),
"Content-Type" = "application/json"
),
body = toJSON(
list(
case_uid = "analysis_001",
profile_uid = "ai_evidence_review_v1"
),
auto_unbox = TRUE
),
encode = "raw"
)
fromJSON(content(res, as = "text", encoding = "UTF-8"))curl "$QB_BASE_URL/api/v1/runs" \
-H "Authorization: Bearer $QB_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"case_uid": "analysis_001",
"profile_uid": "ai_evidence_review_v1"
}'FAQ
Questions before trusting the result.
Is QuantBayes Studio an AI model?
No. It is an evidence verification layer. It can be used beside AI models, agents, scientific pipelines, or human review workflows.
Does it decide whether something is true?
No. It measures evidence sufficiency under a declared rule set. The result remains tied to the evidence profile and checked evidence.
Can it work with ChatGPT, Claude, Gemini, DeepSeek, or local models?
Yes. A model can produce an answer, plan, source list, or analysis. QuantBayes Studio can then verify the evidence record against a profile.
What is a profile?
A profile is a reusable checklist of evidence rules. It defines what must be checked before a result is trusted.
What is saved in a run?
A run saves the profile, evidence matrix, QuantBayes result, parameters, and report. This makes the result inspectable later.
Is this only for AI agents?
No. The QuantBayes engine came from scientific evidence analysis, particularly clinical genetics, but QuantBayes Studio is general. It can be used for AI, research, engineering, clinical, and regulatory workflows.
Developer navigationSensible page orderShow links
A compact internal map for reviewing the current product flow while the Studio interface is still being refined.
Public and account pages
Studio workflow
| Order | Page | Purpose | Link |
|---|---|---|---|
| 1 | Studio | Main project, profile, case, and run workbench | Open |
| 2 | Profile database | Browse and compare evidence profiles | Open |
| 3 | New profile | Build or upload a reusable evidence profile | Open |
| 4 | Demo QEM case | Case evidence table with the selected profile | Open |
| 5 | Newest demo result | Saved result display with distribution plot and credible intervals | Open |