BIQSA
BIQSA AI Council

Behaviour Intelligence & Quantum System Architecture

The Right AI in the Right Place

Operational
A

Mission

MISSION: Raw Idea → AI Discussion → Blueprint → Product

We turn raw ideas into working software by combining expert AI perspectives, structured debate, and a blueprint the Software Factory can execute.

💡
Raw Idea
💬
AI Discussion
📄
Blueprint
📦
Product

v0.7 Council workspace

Council Orchestration

The Right AI in the Right Place

Copy prompt → open your AI → paste the reply → it appears in the Council Feed → open Consensus when ready.

Builder War Room

BIQSA routes each AI seat to the role, round, and source context where it performs best.

Round 1: ProposalRequired 0/3All seats 0/6 receivedCurrent: Architect · ChatGPTRunner: ManualConsent: No ticketPlan: Free · Manual Seat

Round Manager

Generate round prompts · applies to the whole Council

Round-aware Seat Runner v1

Manual Seat Mode · Round-aware Seat Runner coordinates delivery — you send prompts and paste responses yourself.

Not started
Round
Round 1: Proposal
Required
0/3 received
Optional
0/1 received · 0 active
Courier jobs
0
Responses pasted
0
Marked received
0
Who should act next? Researcher · Gemini — Prepare Courier Job

Per-seat runner

  • Architect · ChatGPT

    Architect · ChatGPT · required

    Courier: —Response: Missing→ Prepare Courier Job

  • Evidence Finder · Perplexity

    Evidence Finder · Perplexity · required

    Courier: —Response: Missing→ Prepare Courier Job

  • Researcher · Gemini

    Researcher · Gemini · required

    Next
    Courier: —Response: Missing→ Prepare Courier Job

  • Builder

    Builder · GitHub Copilot · optional

    Courier: —Response: Missing→ Prepare Courier Job

Round behavior

Round 1: Proposal · Builder War Room

Because the pattern is Builder War Room, this round asks seats to name components, apis, and the thinnest viable slice.

ObjectiveDefine build scope, interfaces, and technical constraints.
Seat styleName components, APIs, and the thinnest viable slice.
Challengelight
Evidencecite_when_claiming
ConsensusShip decisions only with owner, dependency list, and next build slice.

Optimize for what can land in the next increment, not theoretical completeness.

Quality gate

Needs responses

Advisory only · 0/3 required · 0 evidence · 0 for consensus

Evidence Recommended · Evidence recommendedEvidence is recommended during early ideation. BIQSA suggests at least one included source in the Evidence Pack.

This round is not ready because 3 required seat(s) have not marked a response received (Architect · ChatGPT, Researcher · Gemini, Evidence Finder · Perplexity).

  • Required seats: 0/3 required responses received
  • Optional seats: 0/1 optional responses received
  • Evidence posture: Evidence Recommended (Evidence recommended) · 0 included · min 0
  • Evidence pack: 0 source(s) — Evidence recommended

AI IQES (advisory)

IQ · Not assessedEQ · Needs reviewSQ · Needs reviewPQ · Not assessedBalance · Needs review

Complements Quality Gate & Evidence Posture — not numeric scoring. Human validation required.

Current seat · actions

Required seats gate round advance · optional adds depth

Selected seat

ChatGPT logo

Architect · ChatGPT

Architect · ChatGPT

This round
Required
Role
Architect
Round
Round 1: Proposal
Prompt
Prompt Ready
Response
Pending
Conversation URL
Not saved
Evidence Pack
0 included

Advance round

Waiting for required seats: Architect · ChatGPT, Researcher · Gemini, Evidence Finder · Perplexity

Prompt Courier v1

Manual Seat Mode is official. Prompt Courier does not send prompts — you copy and paste into your provider thread yourself.

Human approval required before send. BIQSA never auto-sends to a provider. No auto-send, browser automation, or provider API calls.

View Prompt
# BIQSA AI Council Prompt
## Session
Title: BIQSA Council Session
Council mode: Build
Desired output: Product Blueprint
## Communication pattern
Builder War Room
## Round behavior
Pattern Builder War Room · Round 1: Proposal behavior
Objective: Define build scope, interfaces, and technical constraints.
Style: Name components, APIs, and the thinnest viable slice.
Challenge light · Evidence cite_when_claiming
Consensus rule: Ship decisions only with owner, dependency list, and next build slice.
Optimize for what can land in the next increment, not theoretical completeness.
## Mission Brief
(mission not provided)
## Council mode guidance
Council mode: Build — Implementation focus — architecture, APIs, data, UX flows, and delivery risks.
Prompt depth: balanced. Consensus bar: normal.
Favor roles (required bias): architect, engineer, product.
Supporting roles (optional bias): qa, devops.
Recommended rounds for this mode: proposal → improvement → consensus → notulen.
Include components, interfaces, data model notes, and implementation sequencing.
## Source Pack
No source materials added yet.
## Source grounding
No sources are currently included in the Evidence Pack. Build mode: tie requirements to implementation choices (stack, APIs, data, UX, milestones).
Round 1 (Proposal): Use the Source Pack to understand product, company, user intent, constraints, and the stated output goal.
Do not invent unsupported claims. If sources conflict, mark it as an open question.
## Your Seat
Architect · ChatGPT
Seat role: Architect
Assigned provider: ChatGPT
Access mode: manual
Output expectation: Architecture recommendations and MVP system boundaries.
Notes: Strong for structured proposals and synthesis.
## Current Round
Round 1: Proposal
Intent: Offer the strongest initial answer.
## Previous Council Responses
(no previous responses yet)
## Instructions
Respond as the Architect. Target deliverable type: Product Blueprint.
Keep the answer practical and scannable — enough detail to act on, without filler.
Use Markdown headings and bullets.
For this round, focus on: Offer the strongest initial answer.
End with 2-4 bullet recommendations.
## Response footer (required)
End your response with this footer exactly:

---

Prepared through BIQSA AI Council
https://council.biqsa.app
Session: BIQSA Council Session
Round: Round 1: Proposal
Seat: Architect · ChatGPT
Open Provider

Chat Link / Conversation URL

Round Chatroom

Current round focus · 0 received

Sources included: 0 · No sources yet

Current seat round: Round 1: Proposal

No responses in this round yet. Start with the required speakers below.

Required speakers

  • ChatGPT logo

    Architect · ChatGPT

    Architect

    Waiting

    Continue
  • Gemini logo

    Researcher · Gemini

    Researcher

    Waiting

  • Perplexity logo

    Evidence Finder · Perplexity

    Evidence Finder

    Waiting

Optional speakers

  • GitHub Copilot logo

    Builder

    Optional
Not needed this round · 2 seats
  • Claude logoCritic · Claude · Not needed
  • NotebookLM logoNotulen · Not needed

Council room is ready.

Paste the first AI response for this round.

Start here

  1. Fill mission and pick your AI seats
  2. Copy the prompt (View Prompt if you need to check it)
  3. Open your AI provider and run the prompt
  4. Paste the reply — it appears in the Council Feed
  5. Mark Received, then open Consensus when you are ready

1. Council Session Setup

How this works: Mode controls how BIQSA runs the council. Desired Output controls what BIQSA produces.

Implementation focus — architecture, APIs, data, UX flows, and delivery risks.

Build mode works best with sources in the Evidence Pack. Consider adding evidence before copy/paste rounds.

Council Decision Pack

Technical

Meeting-ready outputs from evidence plus AI deliberation — decisions, actions, and client summary.

Advisory draft

This round is not ready because 3 required seat(s) have not marked a response received (Architect · ChatGPT, Researcher · Gemini, Evidence Finder · Perplexity).

Quality Gate: Needs responses · Evidence Recommended (Evidence recommended) · 0 source(s) in pack

Implementation detail — assumptions, dependencies, and operational constraints.

Output Style shapes Decision Pack presentation only. Evidence Posture, Quality Gate, AI IQES, cyber safety wording, and human validation requirements are unchanged.

AI IQES (advisory)

IQ · Not assessedEQ · Needs reviewSQ · Needs reviewPQ · Not assessedBalance · Needs review

Complements Quality Gate & Evidence Posture — not numeric scoring. Human validation required.

Output / Session Radar

v1-taxonomy-patch · output/session posture only

Evaluates whether council outputs and the session as a whole are logical, context-fit, responsible, and executable — not whether an AI model is qualified for a seat role.

Draft / advisory / provisional — not canonical IQES. Draft / advisory / provisional — not canonical IQES. Measures Output/Session quality posture of council material, not AI provider capability or agent runtime safety. Human review and Evidence Posture / Quality Gate remain authoritative. No automation is triggered from these scores.

IQES layer taxonomy & RACI

**IQES Radar taxonomy (MindFactory v2):** - **Output IQES** — one answer, proposal, artifact, or result (this chart’s dimensions at output scope). - **Session IQES** — whole council session completeness, dissent, evidence, and decision basis (session view). - **Seat Capability IQES** _(separate panel — Seat Capability Radar)_ — whether an AI/model/agent is qualified for a seat role. **Not this chart.** - **Agent Runtime IQES** _(separate panel — Agent Runtime Radar)_ — safe behavior under permission in audit records. **Not this chart.**

**IQES** defines capability dimensions. **RACI** defines responsibility (who is Accountable/Responsible). **Authorization** defines permission to act. Do not use this Output/Session Radar to decide provider qualification or runtime safety.

Scoring: 0/100 = evaluated and failed. N/A = not enough evidence (not a failure).

Pre-Council — add a mission brief and seat responses before Output/Session radar can assess quality posture.

IQPre-CouncilEQPre-CouncilSQPre-CouncilPQPre-Council
Balance gap
N/A
Evidence confidence
insufficient
Overall
Pre-Council

Dimension scores

IQPre-CouncilN/A

Pre-Council — add a mission brief and seat responses before Output/Session radar can assess quality posture.

EQPre-CouncilN/A

Pre-Council — add a mission brief and seat responses before Output/Session radar can assess quality posture.

SQPre-CouncilN/A

Pre-Council — add a mission brief and seat responses before Output/Session radar can assess quality posture.

PQPre-CouncilN/A

Pre-Council — add a mission brief and seat responses before Output/Session radar can assess quality posture.

Blockers / SQ overrides

  • Quality Gate: Needs responses — treat outputs as advisory draft.
  • Output/Session Radar: insufficient council output — not a failure score.
AI IQES Review

AI IQES complements Evidence Posture and Quality Gate. Evidence grounding and procedural gates still apply. IQES does not certify correctness, ethics, or feasibility on its own.

IQ — Reasoning Quality

Not assessed

Logic, analysis, correctness

Reasoning is not yet reliable enough for high-stakes use — collect stronger council input first.

  • Mission brief is thin — reasoning may drift without a clear objective.
  • No consensus-eligible responses yet — IQ cannot be fully assessed.
  • Quality Gate: Needs responses — reasoning chain incomplete for this round.

EQ — Human / Context Sensitivity

Needs review

Empathy, stakeholders, communication

EQ needs strengthening — add stakeholder framing, change impact, or communication sensitivity.

  • Consider whether recommendations account for people affected — mission brief is technical/abstract.

SQ — Values / Responsibility Alignment

Needs review

Ethics, direction, humility, meaning

SQ needs review — authorization, validation, or advisory posture gaps increase ethical delivery risk.

  • BIQSA treats value alignment as responsibility, humility, and meaning — not mysticism.
  • Pack is advisory draft — SQ demands humility; do not overstate certainty to clients.

PQ — Practical Executability

Not assessed

Operational viability, real-world constraints

Mostly executable direction — tighten Action Plan with owners, dependencies, and evidence.

  • PQ means practical executability and operational viability — not physical wellness.
  • Desired output (Product Blueprint) sets a practical delivery target.
  • Round focus: Define build scope, interfaces, and technical constraints.

Overall IQES Balance

Needs review

Needs holistic review across IQ / EQ / SQ / PQ

Weakest lens: Needs review. Quality Gate (Needs responses) and Evidence Posture still govern evidence sufficiency — IQES does not override them. Current gate state suggests holding external commitments until procedural gaps close.

Human Reviewer Notes
## Human Reviewer Notes

Complete before treating council output as client-ready. IQES is advisory — it does not replace Evidence Posture, Quality Gate, or human sign-off.

- [ ] Material claims checked against the Evidence Pack
- [ ] **IQ** — reasoning is logically sound; gaps flagged or accepted with rationale
- [ ] **EQ** — stakeholders and human context respected in tone and recommendations
- [ ] **SQ** — values, responsibility, and humility; no overclaiming or false certainty
- [ ] **PQ** — recommendations are operationally viable within real constraints
- [ ] Cyber/defensive posture upheld where applicable (authorization, read-only, human validation)

**Reviewer name:** _______________________

**Date:** _______________________

**Outcome:** ☐ Approved for discussion  ☐ Revise with council  ☐ Hold for more evidence
Council Minutes
# Council Minutes — BIQSA Council Session

**Session:** BIQSA Council Session
**Mission:** (not provided)
**Desired output:** Product Blueprint
**Council mode:** Build
**Communication pattern:** Builder War Room
**Current round:** Round 1: Proposal — Define build scope, interfaces, and technical constraints.
**Round behavior:** Because the pattern is Builder War Room, this round asks seats to name components, apis, and the thinnest viable slice.

## Deliberation record
_No responses marked received yet._
Executive Brief
_Technical lens — connect recommendations to implementation assumptions, interfaces, and delivery constraints._

# Executive Brief — BIQSA Council Session
## Objective
_Mission brief not set._
## Target deliverable
Product Blueprint
## Council synthesis
No council responses have been marked received yet. When responses arrive, apply Builder War Room consensus rule: Every commitment needs owner + dependency + definition of done.
## Key agreements
- No common agreements extracted yet.
## Caveat
Brief derived from AI council deliberation. Validate against Evidence Pack and human review before external use.
Evidence Map
_Trace requirements and risks to supporting artifacts in the Evidence Pack._

# Evidence Map

**Included in Evidence Pack:** 0 / 0 source(s)

## Source index
_No sources uploaded. Add materials in Sources / Evidence Pack._
Decision Log
# Decision Log

**Communication pattern:** Builder War Room
**Consensus rule (consensus round):** Every commitment needs owner + dependency + definition of done.

## Recorded decisions
- Communication pattern: Builder War Room.
- Consensus rule: Every commitment needs owner + dependency + definition of done.
- Mode selected: Build.
- Active seats: ChatGPT (Architect), Claude (Critic), Gemini (Researcher), Perplexity (Evidence Finder), GitHub Copilot (Builder), NotebookLM (Notulen).
- Decision pending: collect more AI responses.
- No explicit conflict captured yet.

## Agreements
- No common agreements extracted yet.

## Disagreements / tradeoffs
- No explicit disagreements extracted yet.
Action Plan
_List implementation steps with owners, dependencies, and verification criteria where known._

# Action Plan

_Candidate actions from council responses — assign owners and dates in your PM tool._

## From council deliberation

- _No explicit action lines extracted — review Council Minutes._
Open Questions
# Open Questions

## From council responses
- No open questions extracted yet.

## From evidence gaps
- _No aggregated missing-info flags._
Risks / Assumptions
# Risks / Assumptions

## Risks
- Risk log pending more responses.

## Assumptions & limitations
- Council outputs are AI-assisted and require human validation for high-stakes decisions.
- Evidence posture for this session: Evidence Recommended — Evidence is recommended during early ideation. BIQSA suggests at least one included source in the Evidence Pack.
- Absence of a listed risk does not imply the risk is mitigated.
Client-ready Summary
_Client summary (technical) — include assumptions, dependencies, and what must be true for execution._

# Client-ready Summary — BIQSA Council Session
_Advisory draft — complete evidence and quality gate before client delivery._
## What we reviewed
Mission not specified.
## Outcome
No council responses have been marked received yet. When responses arrive, apply Builder War Room consensus rule: Every commitment needs owner + dependency + definition of done.
## Decisions & alignment
- No common agreements extracted yet.
## Risks to acknowledge
- Risk log pending more responses.
## Open questions
- No open questions extracted yet.
_Not a binding commitment — validate with stakeholders and evidence before external distribution._

Shipping focus — architecture, APIs, sequencing, blockers, and factory-ready specs.

BIQSA recommendation: BIQSA selected Builder War Room because Council mode is Build; Desired output is Product Blueprint.

AI Council Planner / Capability Router

The Right AI in the Right Place

Brief + Desired Output + Mode + Sources → recommended AI Council composition.

Sources

No sources yet

AI Seat Summary

Current round: Round 1: Proposal

Active seats
6
Required this round
3
Optional this round
1
Not needed this round
2

Opens More tools → AI Seat Manager.

More tools

Consensus uses Received responses from required and optional seats. Not needed / Extra replies only count if you mark Use in Consensus on the card. Pool: 0 responses (0 total received).

6. Conflict & Consensus

Because the pattern is Builder War Room, this round asks seats to be implementation-specific: apis, data, sequencing, blockers, and acceptance checks.

Key disagreements

  • No explicit disagreements extracted yet.

Common agreements

  • No common agreements extracted yet.

Decision log

  • Communication pattern: Builder War Room.
  • Consensus rule: Every commitment needs owner + dependency + definition of done.
  • Mode selected: Build.
  • Active seats: ChatGPT (Architect), Claude (Critic), Gemini (Researcher), Perplexity (Evidence Finder), GitHub Copilot (Builder), NotebookLM (Notulen).
  • Decision pending: collect more AI responses.
  • No explicit conflict captured yet.

Open questions

  • No open questions extracted yet.

Consensus summary

No council responses have been marked received yet. When responses arrive, apply Builder War Room consensus rule: Every commitment needs owner + dependency + definition of done.