AI in the Design Stage

A I P resentation Three practical w ays A I helps us w ork faster, raise quality and sharpen design thinking.

W hy A I in the design stage 01 E fficiency Less tim e reading, draw ing repetitive options and w aiting for renders 02 Q uality E very requirem ent traced to a clause, few er surprises at review 03 D esign thinking M ore options tested earlier, so decisions rest on evidence Today: three applications · C laude, C hatG P T and P erplexity · tw o live dem onstrations.

Three m om ents in our design w orkflow 01 B rief R egulation R esearch C laude 02 M assing S ite analysis Layout & sketches Form a · C laude · C hatG P T · P erplexity 03 C oncept visualisation x figura · G em ini · C hatG P T.

01 B rief R egulation R esearch C laude turns client briefs and authority regulations into a clear requirem ents and com pliance m atrix W hat it does R eads long briefs, planning regulations and design guidelines in one go E xtracts every requirem ent into a single com pliance m atrix w ith clause references Flags conflicts and gaps betw een the brief and the regulations R eview s a design narrative or area schedule against the brief A i Tool.

01 · H ow it w orks Input D ocum ents C lient brief, authority planning regulations, design guidelines (P D F) C laude O ne prom pt R eads all docum ents together and cross-checks each requirem ent O utput M atrix R equirem ents, clause references, conflicts, gaps and client questions E xam ple prom pt “R ead the attached client brief and the [authority] planning regulations. B uild a com pliance m atrix: requirem ent, brief value, regulation value, clause reference, status. Flag every conflict and gap, and list the questions w e should raise w ith the client.”.

01 · The output: a com pliance m atrix P lot 3450800 · developm ent brief cross-checked against the affection plan and D M parking rule Item B rief asks C hecked against S tatus M ax G FA 42,761 m ² total A ffection plan: 41,479.11 m ² residential + 1,282.06 m ² retail C om pliant H eight G +4P +30, no extra technical floor Floor-count changes w ithin the sam e G FA need an exception approval C heck Tow er setbacks Tow er inside the envelope 5 m on sides 1 and 3 · 10 m on sides 2 and 4 · podium 0 m C om pliant E fficiency 95% N S A /G FA 53% N S A /B U A w ith 50% G FA /B U A im plies N S A ≈ 106% of G FA C onflict P arking ~400+ bays incl. A rada extras D M rule: 1 bay ≤ 150 m ², 2 above · ~372 residential + 17 retail C heck S ky palace 15,750 ft² (~1,460 m ²) Typical tow er floor ≈ 1,380 m ² (41,479 m ² ÷ 30) C onflict.

02 M assing S ite analysis G enerative tools test layouts, sun, daylight, w ind and yield in m inutes, not days. W hat it does G enerates m assing and layout options from plot, setbacks and height lim its R uns sun hours, daylight, w ind and noise analysis on each option C om pares yield, G FA and unit m ix side by side E xports chosen options to R evit or R hino for developm ent A i Tool P erplexity.

02 · W hich tool does w hat A utodesk Form a M assing options inside plot lim its, setbacks and height S un, daylight, w ind and noise studies G FA and yield com parison E xport to R evit or R hino C laude W rites P ython for G rasshopper or D ynam o to autom ate options Turns option data into an area schedule and sum m ary C hecks each option against the 01 com pliance m atrix C hatG P T Q uick layout diagram s from a site plan S ketch ideation for plan arrangem ents First-pass m assing im ages for discussion.

03 C oncept visualisation A I rendering turns rough m assing into m ood im ages for quick option review s W hat it does Turns sketches, m assing screenshots or plans into atm ospheric im ages Tests m aterials, lighting and tim e of day in m inutes P roduces several directions for early internal and client review s K eeps one view point across options so they com pare fairly A i Tool x figura G em ini C hatG P T G endo.

Live exam ple 01 D em onstration S eih S deirah | B icester V illage — one sketch, three A I tools, tw o m assing options.

S eih S deirah | B icester V illage · Input The input O ne rough sketch P lot lim its on both sides of S heikh M oham m ed B in R ashid A l M aktoum S t, a bridge linking the tw o sides, and a lagoon at the heart of the schem e. S am e sketch, sam e prom pt, three tools..

S eih S deirah | B icester V illage · Three tools com pared x figura Follow s the sketch closely; bright, clear daytim e aerial G em ini C lose, dram atic view w ith the bridge as the hero C hatG P T W idest context; m ost natural desert setting.

S eih S deirah | B icester V illage · M assing options O ption A C lustered around the lagoon O ption B Linear spine, bridge across the w ater.

S eih S deirah | B icester V illage · O ption A.

S eih S deirah | B icester V illage · O ption B.

S eih S deirah | B icester V illage · W aterfront.

FitnG lam A bu D habi | B aw adi · E xisting facade E xisting: a long m etal-clad box w ith one entrance block — the base view for the A I facade studies that follow.

FitnG lam A bu D habi | P roposed facade Tw o facade concepts, gradient and band, each tested by day and by night from one base view.

D ow ntow n Tow er II | P roposed facade Facade, podium and roof-terrace view s generated for early concept review.

M ajlis-G reenhouse | S ky M ajlis.

Live exam ple 02 D em onstration D O W N TO W N TO W E R II.

P lot 3450800 | B urj K halifa B oulevard · W ho m ade w hat S am e brief, three tools, three bodies of w ork P art 1 · 6 slides C laude D evelopm ent brief and com pliance m atrix S ite studies and concept m oves on the C A D plot Four options scored, preferred test fit P art 2 · 4 slides P erplexity S ite analysis and tow er concept boards Three option boards: bar, hinge, three-w ing Typical floor test fits and renders P art 3 · 2 slides C hatG P T W atercolour footprint studies and paired volum es C oncept renders in the D ow ntow n skyline.

P lot 3450800 | C laude · D evelopm ent brief The brief in six num bers 42,761 m ² M ax G FA · 41,479 m ² residential + 1,282 m ² retail G +4P +30 M ax height, no basem ent · about 134 m 5,599 m ² P lot area from C A D · brief scaled ~5,400 m ² ~39,400 m ² R esidential N S A at the 95% N S A /G FA target 200–225 E stim ated units, about 7 per floor ~400+ P arking bays incl. A rada extras, in 4 podium levels.

P lot 3450800 | C laude · D evelopm ent brief Four questions raised before any draw ing 01 R atios conflict 53% N S A /B U A w ith 50% G FA /B U A im plies N S A ≈ 106% of G FA , against the 95% target. A re balconies in N S A ? 02 2% contingency D esign to 97% , or hold a buffer at about 93% ? R oughly 9,000 ft² of N S A depends on the answ er. 03 P arking threshold The 150 m ² test is on unit G FA . If brief sizes include balconies, som e 2 B R + M m ay drop to one bay. 04 S ky palace 15,750 ft² (~1,460 m ²) exceeds a ~1,380 m ² typical floor, so it is likely a duplex or terrace-inclusive..

P lot 3450800 | C laude · S ite studies U rban context Location, m ovem ent, open space, public transport and noise S ite and environm ent S un path, W N W w ind, w inter shadow , view s, setbacks and access.

P lot 3450800 | C laude · C oncept m oves S ix m oves from site to form S et the envelope P odium fills the plot Tow er placed for view s C lose the w est, shade the south D eck on the sheltered south-east R etail corner, one access E ach m ove answ ers a site finding: the view axis, the w est sun, the sham al w ind and the single access road..

P lot 3450800 | C laude · M assing options on the C A D plot Four options, scored against the brief O ption P late G FA , 30 fl. B urj face W est face H / W C orridor S core A · P oint tow er 1,623 m ² 41,479 m ² 25% 25% 3.3 9 m 42 · 1st D · A rc slab 1,592 m ² 41,479 m ² 33% 15% 4.8 24 m 39 · 2nd C · B urj slab 1,592 m ² 41,479 m ² 35% 15% 5.2 24 m 38 · 3rd B · B oulevard slab 1,502 m ² 38,849 m ² 0% 35% 5.2 22 m 25 · 4th.

P lot 3450800 | C laude · P referred option A · Test fit 217 U nits · brief asks 200–225 38,877 m ² N S A · 99% of target 40,923 m ² R esidential G FA used 414 / 604 P arking bays required / capacity A E D 2.53bn Indicative G D V at brief rates.

P lot 3450800 | B urj K halifa B oulevard · W ho m ade w hat S am e brief, three tools, three bodies of w ork P art 1 · 6 slides C laude D evelopm ent brief and com pliance m atrix S ite studies and concept m oves on the C A D plot Four options scored, preferred test fit P art 2 · 4 slides P erplexity S ite analysis and tow er concept boards Three option boards: bar, hinge, three-w ing Typical floor test fits and renders P art 3 · 2 slides C hatG P T W atercolour footprint studies and paired volum es C oncept renders in the D ow ntow n skyline.

P lot 3450800 | P erplexity · S ite analysis and tow er concept S ite analysis R oads, access, urban grain, greenery, noise and view s R esidential tow er concept A rea split, vertical program m e, setbacks, sun and w ind.

P lot 3450800 | P erplexity · M assing option boards A · S lender view bar R egular grid, sim ple stacking B · G arden hinge P rotected am enity court, corner hom es C · Three-w ing landm ark M ultiple outlooks, shorter corridors C om m on basis: ≈1,383 m ² plate × 30 floors = 41,479 m ² residential G FA , on a plot traced and scaled to ~5,400 m ².

P lot 3450800 | P erplexity · Typical floor test fits A · S lender view bar 7 apartm ents · 1,077 m ² apartm ent zones B · G arden hinge 7 apartm ents · 1,010 m ² apartm ent zones C · Three-w ing 6 apartm ents · 886 m ² apartm ent zones W hat the test fits show ed: none of the three proved 95% N S A or the 200–225 unit target; C laude's point tow er reached 217 units at 99% ..

P lot 3450800 | P erplexity · C oncept renders A · S lender view bar B · G arden hinge C · Three-w ing landm ark.

P lot 3450800 | B urj K halifa B oulevard · W ho m ade w hat S am e brief, three tools, three bodies of w ork P art 1 · 6 slides C laude D evelopm ent brief and com pliance m atrix S ite studies and concept m oves on the C A D plot Four options scored, preferred test fit P art 2 · 4 slides P erplexity S ite analysis and tow er concept boards Three option boards: bar, hinge, three-w ing Typical floor test fits and renders P art 3 · 2 slides C hatG P T W atercolour footprint studies and paired volum es C oncept renders in the D ow ntow n skyline.

P lot 3450800 | C hatG P T · M assing and footprint studies C om pact, slab and cham fered tow ers P late sizes, cores and podium am enities at 1,383 m ² per floor D · P aired volum es Tw o tow ers on a shared podium , 12 m study gap.

P lot 3450800 | C hatG P T · C oncept renders C om pact tow er C ham fered tow er C urved slab.

U sing A I w ell 01 V erify C heck every clause against the source docum ent before issuing 02 Judge A I im ages show m ood; confirm scale, setbacks and heights in our m odels 03 P rotect K eep confidential client data in approved tools only 04 S hare S ave prom pts that w ork in a shared team library.

Tool com parison · P lot 3450800 A ccuracy: how far can w e trust it? C riterion C laude P erplexity C hatG P T S ite geom etry S trong R eal C A D plot, 5,599 m ², north set P artial Traced from aerial, scaled to ~5,400 m ² W eak Illustrative plot outline B rief checking S trong 4 conflicts found, 8-criteria scoring P artial C ontrols transcribed, gaps flagged W eak B rief figures restated, not checked Y ield vs targets S trong 217 units, N S A 99% of target W eak O w n notes: 95% N S A not show n W eak H om es and parking unverified A rithm etic S trong A reas, units and bays tie up S trong S am e plate and G FA throughout S trong P lates add up to 1,383 m ² Team review of the outputs in this deck · one project, schem atic stage · all three still need survey, structure and fire checks.

Tool com parison · P lot 3450800 Q uality: how w ell does it present? C riterion C laude P erplexity C hatG P T D iagram s P artial C lear technical boards, dense S trong C lean colour boards, easy to read S trong A ttractive w atercolour studies Floor plans P artial U nit blocks, m ix vs brief S trong R oom -level layouts and schedules W eak N o floor plans R enders N one N o im ages produced S trong P hotoreal, podium follow s the plot P artial P hotoreal, podium often sim plified C lient-ready P artial N eeds a visual layer S trong B oards and im ages in one style S trong B est m ood and atm osphere Team review of the outputs in this deck · one project, schem atic stage.

Tool com parison · V erdict U se each tool for w hat it does best C laude M ost accurate B rief and regulation checks, option scoring, area and unit num bers w e can verify P erplexity B est all-round visuals R eadable site boards, room -level floor plans and renders that follow the plot C hatG P T B est atm osphere Q uick sketch studies and m ood im ages for early client conversations Together: C laude's num bers, P erplexity's plans, C hatG P T's im ages. N one of them replaces our survey, structure, fire and area checks..

Thank you Q uestions and discussion E N D.