Brainstorms, retrospectives, structured comparisons — each one a named technique with literature behind it, structured for the kind of question it answers.
What each methodology returns
Every methodology takes a question, routes it to the right people and agents, and returns a typed answer your team — or your code — can act on. Conflicts get flagged, outliers stay surfaced, attribution stays honest.
Synthesis · Brainstorm
12/12 responded
Q4 pricing tier — themes
Outliers · 2 · See list →
Divergent
Generate ideas where none yet exist. The point is breadth.
Brainstorm
Open generation, clustered into themes.
Free-form idea generation from each responder. Confer clusters responses by theme, labels each cluster, and surfaces outliers that differ meaningfully from the rest.
Synthesis
Themes with member counts, representative responses, outliers.
Example
Question
What should we name the new pricing tier?
Audience
Marketing, sales, and product (12 people)
Response
Free text per responder
Synthesis
Clusters of name themes, with outliers and rationale per cluster.
Convergent
Collect structured input and reconcile it into one answer. The point is to land on something usable.
Info Gathering
Structured fields per responder, reconciled per field.
Ask specific factual or opinion questions. Each responder fills a short structured form; Confer reconciles answers per field, flags conflicts, and surfaces per-field attribution.
Synthesis
Per-field reconciled answer, attribution, conflict markers.
Example
Question
Q4 hiring priorities by team
Audience
Engineering leads (5 people)
Response
Typed fields — number, text, single-select
Synthesis
One reconciled value per field, attribution, and conflict flags where responders disagreed.
Adaptive Survey
Authored questions plus model-generated follow-ups.
Designer-authored initial questions with LLM-generated follow-ups based on each respondent's answers and on themes emerging across the group as responses arrive. Runs fully anonymous — useful when honest signal matters more than attribution.
Synthesis
Per-question reconciled themes, follow-up-driven depth, surfaced emergent threads, anonymized contributor coverage.
Estimation
Numeric forecasts with reasoning, distributed.
Each responder submits a numeric value and a one-line reasoning; Confer returns a distribution (mean / median / stddev / p10 / p50 / p90), reasoning clusters, outliers, and a spread assessment.
Synthesis
Distribution and central tendency, reasoning clusters, outliers, spread assessment.
Example
Question
What revenue should we forecast for Q3?
Audience
Finance + GTM (8 people)
Response
One number + a single-line rationale
Synthesis
Median, p10/p90, the strongest reasoning clusters, and the outlier worth a second look.
Pros & Cons
One yes/no proposition, structured argument.
Evaluate a single yes/no proposition. Each responder lists their pros and cons and an optional leaning; Confer groups similar points across responders, weights them by recurrence and framing, and returns a net assessment with the strongest pro and strongest con called out.
Synthesis
Grouped pros and cons with supporter counts, weighted net leaning, the strongest pro, the strongest con.
Compare
What you like and don't, per option.
Qualitative comparison of 2–5 options. You supply the options; each responder writes what they like and what they don't like about each one. Confer groups similar points across responders, weights them by recurrence and framing, gives each option a net assessment, and (when one option clearly leads) recommends a winner.
Synthesis
Per-option likes and dislikes with weights and supporter quotes, a per-option net assessment, and a recommendation when one option clearly leads.
Reflective
Bounded look-back producing learning. The point is to extract durable lessons from a completed period.
Retrospective
Bounded look-back, clustered into actions.
Look back at a bounded period of work. The team reflects on a familiar template (Start / Stop / Continue and others) using a Kanban-style card form, the LLM groups similar cards, and the synthesis produces up to three action suggestions grounded in cluster summaries and any prior action suggestions you carry in.
Synthesis
Per-column clusters with supporters, cross-column patterns, up to three grounded action suggestions.
Example
Question
Sprint 24 retrospective — Start / Stop / Continue.
Audience
Squad of 8 engineers + PM
Response
Cards per column, one starter required, free Add card form
Synthesis
Clusters per column with citation counts and three suggested actions, each grounded in a motivating cluster or prior memory.
Prioritization
Allocate finite attention or resources across known options. The point is allocation, not truth.
Prioritize
Rank a list under a stated criterion.
Rank a list of N items under a stated criterion. Each responder rates every item 1–5 and picks their top 3; Confer ranks the list, splits into tiers, and flags contested items and drop candidates.
Synthesis
Ranked list with tiers, per-item supporter counts, contested items, drop candidates.
Single ask
Skip the synthesis layer. One question, one specific person, a typed answer back. The point is to land an answer fast when synthesis would only get in the way.
Structured Ask
One question, one specific person, typed answer back.
Pick a response shape — yes/no with reason, number, short text, long text — and send the question to one specific person. No group required: the person record is created on the first ask and reused on every subsequent one. Confer skips synthesis and returns the typed singular answer.
Synthesis
Typed singular answer per response shape, attribution, submission timestamp.
Example
Question
Should we treat the Series B SAFE as equity for the Q3 close?
Audience
One person — your CFO
Response
Yes / no with a one-line reason
Synthesis
{ answer: "yes", reason: "Conversion is mandatory at the next priced round; matches the auditor's prior-year treatment." }
Expert agents
Every methodology can include hosted agents alongside the humans. Confer ships four archetypes — each calibrated for a distinct role and grounded in a fact-list knowledge base you populate. Drop one or several into any group; methodologies treat them like humans.
institutional_memory
Institutional Memory
What your org has decided, and what came of it.
Carries the past decisions, outcomes, and commitments you indexed. Refuses to extrapolate beyond known facts; cites dates when its facts have them. The voice of "here's what we tried in Q2 2024 and how it landed."
position_taker
Position Taker
A defined stance, argued from the lens.
Holds a calibrated point of view — a CFO lens, a security lens, a GTM lens. Argues from that vantage and surfaces the tradeoffs the stance cares about. Doesn't false-balance into neutrality.
domain_expert
Domain Expert
Neutral specialist, evidence-grounded.
A bounded specialist on a topic — pricing, EU regulation, payments rails. Answers from its indexed knowledge base; explicitly defers when a question lands outside what its facts cover. Supplies what the evidence supports.
persona
Persona
Reasoning in someone's voice.
Embodies a specific person — real or composite. Reaches for the angles they would reach for; uses their phrasing and cadence; declines in their voice rather than breaking character. Useful for boards-of-advisors, customer composites, founder archives.
Pick a methodology. Send a question. Get one answer back.
Sign in, add a few people, mix in archetype agents, and run any methodology — from a five-minute brainstorm to a sprint retrospective to a draft review.
Free during preview · No credit card required