DREAMVISION AI Private briefing ↗
CREATIVE INTELLIGENCE INFRASTRUCTURE

Creative Intelligence,
encoded.

The taste and judgement layer for creative teams.

DREAMVISION turns a practice’s project history and senior judgement into private operational intelligence, pluggable into the tools, workflows and agents the team already uses.

Book a private briefing ↗ See the system ↓
DREAMVISION AI creative knowledge landscape
THE RECORD Projects become evidence.
Decisions become lineage.
THE MISSING LAYER

Generation is no longer the bottleneck.
Judgement is.

Studios have more models, more data and more automation than ever. The hard part still lives in the quiet decisions: what belongs, what is weak, what to preserve, what to remove, and when to escalate. Perception, quality control and critical thinking.

TOOLSModels · BIM · CAD · automation
MISSINGTaste · critical judgement · institutional memory
DREAMVISION AIOperationalises the second inside the first.
FROM THE FOUNDER
Generative tools multiplied everything but they never replaced taste, instinct, and critical thinking.
ILA COLOMBO · FOUNDER, DREAMVISION AI
HOUSE GENOME

The practice, encoded around its own ontology.

DREAMVISION does not train a generic model on a studio archive. We convert authorised material into evidence objects, relate them through a domain ontology, then validate which relationships become institutional judgement.

House Genome A private, versioned representation of what the practice knows, values, protects and refuses.
01 / AUTHORISED EVIDENCE 02 / DREAMVISION AI ONTOLOGY 03 / GOVERNED INTELLIGENCE PROJECT RECORD models · drawings · renders · metadata REVIEW RECORD comments · revisions · approvals · rejections REFERENCE RECORD precedents · materials · visual cues · briefs SITUATED REASONING voice · place · image · live project context PROJECT CONTEXT REFERENCE DECISION RATIONALE REFUSAL THRESHOLD PERSON / ROLE HOUSE GENOME institutional baseline · validated studio logic LEAD GENOMES authorised senior judgement · person-specific logic LIVE PROJECT CONTEXT the decision state currently in front of the team observed evidence provisional relation human-validated judgement
01 Evidence is preserved with source, version and permissions. 02 DREAMVISION frames relationships through its own ontology. 03 Only validated reasoning becomes governed intelligence.
Creative team reviewing work together
Capture the reasoning while it still belongs to the people who hold it.
LEAD GENOMES

Encode how senior people decide.

A Lead Genome is not a style fingerprint. It captures a named senior’s validated reasoning pattern: what they notice, which precedent they recall, how they evaluate a trade-off, what they reject, and how they direct the next move.

01Review momentslive critique · redlines · option comparisons
02Reasoning episodesvoice · project/site context · decisions as they form
03Counterexampleswhere a preference stops applying
04Senior validationconfirm · qualify · reject · scope
JUDGEMENT TRACE

Repeated decisions become a pattern only after validation.

Each column is a real reasoning episode. The Genome is the versioned pattern across them, never a single inferred preference.

DREAMVISION AI RUNTIME

Judgement where the work already happens.

Ambient by default. Queryable on demand. Plugged into your software. DREAMVISION can evaluate work quietly inside the studio’s existing CAD, BIM, modelling, review or agent workflow, surfacing direction only when it adds value.

The team does not have to leave its existing environment to use DREAMVISION.

Illustrative DREAMVISION AI judgement: floating review notes around a working model inside the studio workspace
ONE GOVERNED RUNTIME
01
Ambient

Evaluate quietly in the background. Surface only meaningful conflict, drift or opportunity.

02
Queryable

Let a person interrogate the House or an authorised Lead Genome when deeper reasoning is useful.

03
Plugged in

Internal agents and software can request judgement themselves, through the Genome API.

IN-TOOL GENOME API SECURE DASHBOARD
CAD · BIM · RHINO · REVIEW ENVIRONMENTS · INTERNAL AGENTS Taste and judgement, pluggable into the existing stack.
WHO BENEFITS MOST

Where the quality of the work depends on creative direction and taste, not just execution.

DREAMVISION is most valuable where expertise is concentrated, archives are deep, and AI transformation is creating more output than the organisation can reliably judge.

Sculptural architecture at dusk: work where creative direction and taste decide the outcome
01Design-led architecture practices

Keep house judgement coherent as teams, tools and project types scale.

02Studios with senior knowledge at risk

Preserve how experienced people evaluate and solve problems before transition or retirement.

03Multidisciplinary firms with fragmented intelligence

Connect project precedent, reviews, references and decision history across silos.

04Digital and AI transformation leaders

Make AI useful inside the existing stack without introducing another disconnected workflow.

FORWARD-DEPLOYED ENCODING

Proven on one workflow.
Grown into infrastructure.

The first deployment is intentionally bounded: one decision surface, a selected evidence set and named senior validators. The goal of our Forward Deployed Engineers implementation is to accurately build and prove how the studio’s own judgement can be made operational. The governed intelligence layer expands from there.

01 FOCUS

Select the decision surface.

Concept review, design QA, site judgement or another workflow where senior reasoning materially changes the outcome.

OUTPUT / first bounded use case
02 ENCODE

Build the first House Genome.

Authorised evidence is normalised, related through the DREAMVISION ontology and reviewed with senior validators.

OUTPUT / governed institutional baseline
03 SHARPEN

Add authorised Lead Genomes.

Capture how selected senior people evaluate, refuse, direct and escalate through repeated validated reasoning episodes.

OUTPUT / person-specific judgement profiles
04 OPERATE

Put judgement into live work.

Release the governed runtime through in-tool surfaces, API calls or secure review environments.

OUTPUT / first operational proof
05 EXPAND

Grow from proof into infrastructure.

Add project families, senior Genomes, teams and workflows as the intelligence layer earns broader use.

OUTPUT / expanding studio intelligence
DEPLOYMENT LOGIC Narrow enough to validate. Useful enough to become infrastructure.
GENOME ZERO / RESEARCH

The rare part was never the output.

Our creator-level research studies how references, refusals, decisions and quiet judgement can become legible without collapsing a person or practice into a style clone.

Genome Zero: the archive mapped, reference and project records dissolving into structured creative signal
RELATIONAL CORPUS

We study the structure between the choices.

Individual files are only evidence. The deeper signal appears in repeated relationships: which references precede a decision, what gets refused, when a threshold changes, and whose judgement carries authority.

The relational corpus: spectral fibres of related creative material flowing through the archive
VALIDATED RELATIONAL FIELD Judgement becomes legible
through repeated structure.
Illustrative review card in the relational corpus: keep, push and refuse annotations across a project image, with references and the record beside it
DREAMVISION AI

Make your creative intelligence operational.

The taste and judgement layer for creative organisations.

Book a private briefing ↗