Method
Built to run
decisions
before they're made.
How Simulatte's Decision Engineering infrastructure works.
Most decisions fail not from bad intent, but from untested assumptions. The real world is expensive to test in. We built an alternative — starting with the people those decisions affect.
01 — How they're built
133 attributes. Built layer by layer.
Each deep persona is the product of an 8-stage pipeline. Real-world data grounds the generation. Progressive conditional filling ensures attributes correlate the way they do in reality — not randomly assigned.
Persona Assembly Pipeline
INITIALISING
Pipeline Manifest
01
Corpus
PENDING
02
Taxonomy
PENDING
03
Anchor
PENDING
04
Generation
PENDING
05
Memory
PENDING
06
Persona Ready
PENDING
01 · Corpus Ingestion
> interview_transcripts/
> survey_open_ends.csv
> market_report_q4.pdf
> brand_tracker_23.xlsx
────────────────────────────────
847 signals extracted ✓
02 · Taxonomy Construction
BASE ATTRIBUTES 100
DOMAIN EXTENSION + 033
─────────────────────────────────
TOTAL 133
100% mapped to ICP spec
03 · Demographic Anchor
NAME → Sarah Mitchell
AGE → 34
LOCATION → Portland, OR
LIFE STG → Parent
EMPLOY → Full-time
──────────────────────────────────
ANCHOR LOCKED ✓
04 · Persona Generation
PG-001 · Sarah Mitchell
34 · Portland, OR · Parent
analytical · authority-trust · low risk
PG-002 · James Harrington
42 · Chicago, IL · Professional
deliberate · peer-trust · medium risk
PG-003 · Elena Voss
29 · Berlin, DE · Early career
intuitive · social-proof · high risk
PG-004 · Marcus Reeves
28 · Austin, TX · Freelancer
impulsive · experience-trust · high risk
Generating cohort
0 / 4
05 · Memory Construction
CORE MEMORY CONSTRUCTING
IDENTITY STATEMENT ✓
KEY VALUES ✓
CONSTRAINTS ✓
RELATIONSHIP MAP ✓
WORKING MEMORY INITIALISED
OBSERVATIONS [ ]
REFLECTIONS [ ]
BRAND MEMORIES [ ]
06 · Cohort Ready
COHORT-20260411 · 4 PERSONAS
ATTRIBUTES 133 EACH ✓
QUALITY GATES G1–G12 ALL PASS ✓
DISTINCTIVENESS0.82 ✓
NARRATIVES 1ST + 3RD ✓
SIMULATION READY ✓
─────────────────────────────────
COHORT BUILT SUCCESSFULLY
02 — How they think
Every persona runs a cognitive loop.
Perceive a stimulus. Retrieve relevant memories. Reflect. Decide. Each step uses a calibrated model. The loop accumulates experience over time — every interaction makes the persona sharper.
Perceive
→
Reflect
→
Decide
↩
LOOP
01 · Perceive
Haiku
Stimulus Intake
The persona encounters a stimulus — an ad, a price, a recommendation. Each signal is parsed and scored for importance.
> New product placement in feed
> Price signal: $24.99 / month
> Expert endorsement in copy
> Certification badge detected
Importance Score
—
→
02 · Reflect
Sonnet
Memory Retrieval & Reflection
When the importance buffer crosses threshold, the persona retrieves related memories, weighs them, and forms a new reflection.
RETRIEVAL SCORE =
α · recency
β · importance
γ · relevance
──────────────────────
equal weights: α = β = γ
α · recency
β · importance
γ · relevance
──────────────────────
equal weights: α = β = γ
Retrieve relevant memories from core + working
Score by recency, importance, relevance
Synthesise into a new reflection
Commit reflection to working memory
→
03 · Decide
Sonnet
Decision Output
The persona evaluates the decision scenario against its full memory state. A reasoning chain fires through its trust pathway.
REASONING CHAIN ───────────────
authority-trust triggered
→ price within ceiling
→ expert endorsement detected
→ decision threshold reached
authority-trust triggered
→ price within ceiling
→ expert endorsement detected
→ decision threshold reached
Decision
YES
Confidence
78% · analytical · authority-trust pathway
03 — What you get
Complete, decision-ready agents.
Every persona is a fully structured record — demographic anchor, 133+ correlated attributes, an immutable Core Memory, a volatile Working Memory, and a complete simulation history. Swipe through the cohort.
PG-001 · Sarah Mitchell
Simulation-Ready
Demographic Anchor
Sarah Mitchell
34 · Female · Portland, OR
Parent · Full-time · Graduate
Parent · Full-time · Graduate
Behavioural Tendencies
Consults expert reviews before committing to subscriptions
Price ceiling: $30/month before reconsidering
Reads ingredient labels; trusts third-party certifications
Core Memory — Immutable
Identity
"I'm a working parent of two. My family's wellbeing comes before brand loyalty — but I need evidence, not marketing."
Backstory
Sarah switched from a national cereal brand after her pediatrician flagged high sugar content. She spent three weekends comparing alternatives, reading clinical studies, and asking other parents in her co-op. She settled on a smaller brand with third-party nutritional certification. She hasn't switched since — but she checks every quarter.
Key Values
family health
value for money
scientific evidence
Non-Negotiables
expert-endorsed
transparent ingredients
budget: $30/mo
Relationship Map
DECISION PARTNERspouse
KEY INFLUENCERSpediatrician, parent co-op
Simulation Output
YES
78% confidence · authority-trust pathway
authority-trust → price-check → expert endorsement → YES
PG-002 · James Harrington
Simulation-Ready
Demographic Anchor
James Harrington
42 · Male · Chicago, IL
Professional · Full-time · MBA
Professional · Full-time · MBA
Behavioural Tendencies
Consults 3+ colleagues before major purchases
Reads industry publications for validation
Compares 4-5 alternatives before committing
Core Memory — Immutable
Identity
"I've built my career on careful decisions. I don't chase trends — I invest in what my network validates."
Backstory
James once championed a SaaS tool at his firm based on a compelling demo. Six months in, the product underdelivered and he spent weeks migrating the team off it. Since then, he runs every vendor decision through at least three trusted peers before signing. His LinkedIn DMs are full of "quick question about X" — he treats his network as a living due-diligence engine.
Key Values
professional reputation
peer consensus
long-term ROI
Non-Negotiables
peer-validated
proven at scale
exit clause present
Relationship Map
DECISION PARTNERCTO, ops lead
KEY INFLUENCERSMBA cohort, industry Slack
Simulation Output
YES
64% confidence · peer-trust pathway
peer-trust → price-value check → moderate confidence → YES
PG-003 · Elena Voss
Simulation-Ready
Demographic Anchor
Elena Voss
29 · Female · Berlin, DE
Early career · Full-time · Postgrad
Early career · Full-time · Postgrad
Behavioural Tendencies
Follows influencer recommendations readily
Impulse trials under €25 without research
Shares purchases on social media for feedback
Core Memory — Immutable
Identity
"If something feels right and people I follow are talking about it, I'll try it. Life's too short for analysis paralysis."
Backstory
Elena moved to Berlin for a postgrad in behavioural economics and never left. She discovered her favourite skincare brand through an Instagram story from a micro-influencer she follows for book recommendations — not beauty. That crossover felt authentic. She now treats her social graph as a curation layer: if three unrelated people mention the same thing in a week, she buys it that weekend.
Key Values
authenticity
social validation
new experiences
Non-Negotiables
feels authentic
socially shareable
under €25 trial
Relationship Map
DECISION PARTNERflatmate, close friends
KEY INFLUENCERSmicro-influencers, Reddit threads
Simulation Output
YES
89% confidence · social-proof pathway
social-proof → low price barrier → high enthusiasm → YES
PG-004 · Marcus Reeves
Simulation-Ready
Demographic Anchor
Marcus Reeves
28 · Male · Austin, TX
Freelancer · Self-employed · Bachelor's
Freelancer · Self-employed · Bachelor's
Behavioural Tendencies
Decides within 24 hours or moves on entirely
Trusts user reviews over expert opinion
Willing to pay premium for brand alignment
Core Memory — Immutable
Identity
"I go with my gut. If the experience matches the vibe, I'm in. Reviews matter more than credentials."
Backstory
Marcus quit his agency job at 25 to freelance. His first client came from a cold DM after he posted a side project on Twitter. That taught him that authenticity converts faster than credentials. He now evaluates every product the same way: does it feel like it was built by someone who actually uses it? If the onboarding feels corporate, he bounces. If the founder's story resonates, he pays full price without blinking.
Key Values
personal freedom
authentic experience
value alignment
Non-Negotiables
real user reviews
founder transparency
no corporate feel
Relationship Map
DECISION PARTNERnone — solo
KEY INFLUENCERSTwitter mutuals, indie makers
Simulation Output
NO
42% confidence · experience-trust pathway
experience-trust → no user reviews found → price resistance → NO
Sarah Mitchell · PG-001
Pew USA — Study 1A
88.7%
Distribution accuracy across 15 Pew American Trends Panel questions. 2.3pp below the 91% human self-consistency ceiling.
Pew India — Study 1B
85.3%
The first system to replicate India's political and cultural landscape at population scale. 22 calibration sprints from a 46.2% baseline.
vs. Average frontier LLM
4.9×
Closer to the human accuracy ceiling than the average large language model on India cultural questions. 10 LLMs tested, 5,878 SHA-256 verified API calls.
How we work
01
Behaviour, not opinion.
The system models what populations will do — not what they say they will do. Survey data is an input to calibration, not the output of a study.
02
Simulation, not speculation.
Every decision variant is run through the population before it is run in the market. Outcomes are probable, not guessed. Every result is traceable.
03
A discipline, not a deliverable.
We don't produce reports. We produce a Decision Engineering practice the organisation can operate repeatedly — and one that improves with every engagement.
Start a conversation.
Reach us at iqbal@simulatte.io
· Run a Decision at simulatte.io/templates