Methodology essays, opinion pieces, and lessons from building research-grade synthetic participants. New entries publish on a regular cadence.
Some synthetic research tools build AI twins of real people. Candor builds participants from evidence about a type of customer, so there is far less to govern.
Synthetic participants run on models tuned to agree. Here is what Candor does about people-pleasing on both sides of the interview, and where it falls short.
A new report lays out where synthetic surveys fall short. Here is how each criticism applies to synthetic interviews, and what Candor does about it.
Synthetic research output can look like data even when it isn't. Six tells that separate real signal from confident filler, and how to check each one.
Teams obsess over which AI model a synthetic research tool uses. It's the wrong question. What you ground the model in matters far more than which model it is.
63% of research teams have no rules for synthetic users. That gap is a risk, not caution. A one-page starter set of defaults any team can adopt.
Synthetic-then-human is one way to use synthetic research, not the only one. Three patterns teams actually run, and how to tell which one fits your question.
The strongest case against synthetic users is worth taking seriously. Most of it argues for grounding and discipline, not for skipping synthetic research.
Most AI persona tools forget everything between sessions. Candor's participants keep six types of memory, so each stays the same person across a study.
The synthetic user and persona platforms you'll meet in 2026, in nine categories by what they do, plus a seven-point framework for judging any of them.
Most products ship on assumptions nobody tested. A framework for product teams: surface them, validate them with synthetic research, rewrite the spec.
Two synthetic participants with the same demographics can still reason completely differently. The difference is personality. Here's how Candor uses OCEAN, calibrated by region and occupation, to shape research-grade reasoning.
Big brands test 10 to 50 concepts per project at $15K to $50K each, and most die in a meeting first. Synthetic-first sequencing fixes the math.
Synthetic concept testing is a strong use case and an easy one to misuse. A decision framework: four conditions where it works and four where it fails.
Real audiences hold distinct types of people. Group clustering finds them in real evidence: one group per evidence segment, each with calibrated trait ranges.
Evidence grounding is what makes a synthetic research output you can audit and defend in a real decision. Here is how the pipeline actually works.
Most tools invent synthetic participants from a description and call it research. Evidence grounding separates a research instrument from a confident guess.
Longer-form writing about synthetic research. Methodology essays explaining how the platform works under the hood, opinion pieces on where the category is heading, and lessons from building research-grade synthetic participants. The journal is one of the ways the methodology behind Candor stays auditable: anyone can read how the system reasons, why specific design choices were made, and what the approach can and can't do.
Ben Yoskovitz, Candor's founder, writes the journal. Entries are drafted with the same evidence-grounded discipline the product applies to synthetic research: precise about what's verified in the codebase, honest about limitations, and explicit about where the methodology stands today versus where it's heading.
On a regular cadence, typically two entries per week. The cadence covers methodology essays (technical walkthroughs of how parts of the pipeline work), point-of-view pieces (Candor's stance on category questions), and shorter notes when something specific is worth saying. Entries are scheduled in advance and go live at their published time.
Help center articles explain how to use Candor: how to set up a study, how to interpret a participant profile, how to run an interview. Journal entries explain the methodology behind Candor and the category point of view: why evidence grounding matters, how group clustering works, where synthetic research is the right tool and where real-customer research is required. Both are public. Both exist to make the underlying methodology inspectable rather than hidden behind a sales conversation.
Start with a free project, or walk through Candor with us first.