Perly Consulting │ Beck Eco

The State of Play

A living index of AI adoption across industries — where established practice meets the bleeding edge
UPDATED DAILY

The AI landscape doesn't move in one direction — it lurches. Some techniques leap from experiment to table stakes in a single quarter; others stall against regulatory walls, technical ceilings, or organisational inertia that no amount of hype can dislodge. Knowing which is which is the hard part. The State of Play cuts through the noise with a rigorously maintained index of AI techniques across every major business domain — classified by maturity, evidenced by real-world adoption, and updated daily so you always know where you stand relative to the field. Stop guessing. Start knowing.

The Daily Dispatch

A daily newsletter distilling the past two weeks of movement in a domain or two — delivered to your inbox while the index updates in the background.

AI Maturity by Domain

Each dot marks the weighted maturity of practices within a domain — hover for a brief summary, click for more detail

DOMAIN
BLEEDING EDGEESTABLISHED

Video generation — short-form

LEADING EDGE

TRAJECTORY

Advancing

AI generation of short-form video content for social media, advertising, and promotional clips. Includes text-to-video and image-to-video generation for brief clips; distinct from long-form generation which produces extended narratives.

OVERVIEW

Short-form AI video generation has crossed into mainstream production deployment across marketing, social content, and e-commerce workflows, with 1M+ YouTube channels now using AI creation tools daily and $18.6B market value. The practice—generating sub-60-second clips from text, images, or reference materials—is a proven alternative to traditional production for high-volume, low-creative-variance content, with 91% cost reduction ($4.5k traditional to $400/min AI-assisted) and 100x per-second advantage. Runway Gen-4.5 and Google Veo 3.1 have achieved production-readiness: human perception studies show 90% of viewers cannot distinguish AI-generated video from real, while named enterprise deployments (20th Century Studios, Netflix Japan, Lionsgate) execute professional VFX and pre-production workflows. Cost economics enable high-frequency production—brands now post 9.5+ times daily across platforms, impossible without AI. Yet the practice reveals sharp economic and governance boundaries. OpenAI's Sora discontinuation (April 26, 2026) documented the constraint explicitly: inference cost ($1M/day) unsustainable against zero monetization from free-tier users, and the core limit is not capability but business model viability. Adoption stratifies by context: marketing and advertising production-ready at scale; enterprise B2B shifting toward human-led production with AI assistance, rejecting pure AI-generated approaches for authenticity and trust; e-commerce validating 80%+ conversion-rate lifts but limited to product-demo use cases. The defining barriers are now economic (unit costs remain fragile at scale), regulatory (copyright litigation unresolved, content provenance standards absent, mandatory AI disclosure now enforced across YouTube, Instagram, TikTok, Google Ads), and contextual (governance constraints limit deployment beyond low-creative-variance marketing).

CURRENT LANDSCAPE

Market stratification has sharpened by use case and business model. Runway ($5.3B valuation after $315M Series E in Feb 2026, $860M total raised) dominates professional deployment with named clients (20th Century Studios, Alien: Romulus VFX; Netflix Japan; Lionsgate, House of David Season 2; Toei Animation, Bandai Namco, Sony Pictures Japan). The platform has transitioned from experimental tool to integrated pre-production infrastructure, with workflows now combining text-to-video, image-to-video, and video-to-video capabilities. HeyGen (bootstrap-funded, $95M ARR, 85K paying customers, 3.2x viral coefficient) leads avatar-focused platforms through freemium accessibility and integrations (Canva 150M MAU, HubSpot, ChatGPT plugin), bifurcating from Synthesia ($146M ARR) by customer segment (creators vs enterprise compliance). Google Veo 3.1 dominates volume (96.4% of generation activity through YouTube bundling) but generates no revenue, establishing a structural problem: free users produce exploratory content, paying users produce marketing briefs. Pika Labs (14.5M users, $135M total funding, $80M Series B) maintains creator positioning with brand partnerships (Balenciaga, Fenty, Vogue), though credit-burn pricing frustrates professional adoption.

Production workflows have shifted to integrated multi-model infrastructure. Brands now deploy cost-optimized stacks combining Runway for character consistency, Kling for physics, Veo for audio sync, and image-to-video APIs for batch scaling. Documented deployments: 340-SKU e-commerce brand achieved 80% conversion-rate lift and 99% cost reduction; Animatic Media's fully AI-produced channel reached 9,900 monthly viewers. Cost structure enables high-frequency production—brands now post 9.5+ times daily across networks, driving shift from hero-asset to 15-30 variant workflows per video. Engagement parity achieved: AI-generated social clips reach 87% of human-content engagement; product demos 75-82%; brand storytelling 61% (projected to converge by 2028).

However, OpenAI's Sora discontinuation (April 26, 2026) exposed critical constraints. Despite technical breakthrough, Sora faced $1M/day inference costs, peak user base of 1M declining to <500K, and $2.1M lifetime revenue—a deployment failure driven by unit economics, not capability gaps. Analysis: free-tier users generate exploratory fantasies with zero conversion; paying users produce operational briefs (marketing, training). The ROI picture remains contested: 95% of corporate deployments fail to sustain returns beyond pilots. Production constraints persist: physics confusion, spatial inconsistency, texture melting, audio artifacts. Critically, enterprise B2B is shifting toward human-led production with AI assistance, explicitly rejecting pure AI-generated approaches. Quote (90 Seconds, 25K+ production platform): "Nothing compares to a real person sharing authentic experience." Regulatory barriers are tightening: YouTube requires AI content labels; Instagram/TikTok mandate disclosure; Google Ads bans deepfakes; UK statute repeals copyright for pure AI-generated output; SCOTUS settled that pure AI-generated content is uncopyrightable. The practice remains bounded: production-ready at scale in cost-optimized marketing; constrained in creative-demanding, regulated, and trust-critical contexts.

TIER HISTORY

ResearchJan-2024 → Jan-2024
Bleeding EdgeJan-2024 → Oct-2024
Leading EdgeOct-2024 → present

EVIDENCE (85)

— The Editorial independent six-week benchmark (May 2026): 240 prompts across narrative, demo, and graphics; measured prompt adherence, temporal consistency, motion realism, generation cost. Sora 2 and Veo 3.1 lead for photorealistic text-to-video; Runway Gen-4.5, Kling 3.0 scored 9.4/10, 9.3/10.

— Runway market dominance: $315M Series E in Feb 2026 at $5.3B valuation; revenue trajectory $121M (2024) to $90M ARR (mid-2025); estimated 300K customers. Market leadership position and enterprise maturity.

— IAB market analysis: U.S. digital video ad spend $80B (2026), up 11% YoY; two-thirds of video buyers activating AI; market stratification by ROI profile and governance constraints.

— Genesys Growth three-platform comparison: peer-reviewed data showing 30-70% workflow productivity gains when AI accounted for minority of work; industry cases cite 80% task improvements; identifies adoption barriers vs hype.

— Market adoption at scale: 78% of marketing teams deploy AI video in at least one campaign per quarter; $18.6B global market (34% CAGR, up from $5.1B in 2023); 67% of AI-generated content underperforms vs human.

— Thicket Trends May 2026 field analysis: Sora 2 dominates cinematic hero clips on TikTok FYP (8-20s, volumetric lighting); Runway Gen-4 dominates professional editing; Veo 3 leads UGC at 87% engagement parity with human content.

— Koro AI case study: 50-SKU fashion brand deployed AI video generation with product-page scraping and AI avatars, generating 50 ads in 48 hours with 80% CVR lift vs traditional production.

— OpenAI's Sora discontinuation (March 24, 2026) documents deployment failure: unsustainable cost structure ($1M/day), limited sustained user engagement (1M→<500K), and counter-creative bias producing homogeneous outputs. Critical negative signal on economics and user adoption.

HISTORY

  • 2024-Q1: Runway Gen-4 launched with character consistency and physics capabilities; OpenAI announced Sora (pre-release); Pika Labs added sound/lip-sync features. Market projections $2.5–25B by 2026–2027. Quality and misinformation concerns documented.
  • 2024-Q2: Runway released Gen-3 Alpha with photorealistic characters and dense temporal control. Pika Labs raised $80M at $470M valuation. Google Veo launched as free bundled alternative, increasing cost competition. Marketing adoption accelerated (25% investment growth planned); production barriers (visual inconsistency, copyright risk, audience skepticism) remained persistent.
  • 2024-Q3: Runway released Gen-3 Alpha Turbo API for developer integration; added video-to-video transformation feature. Pika 2.2 reached 3.5M monthly visits with 37.3% traffic growth through UX improvements. Adobe Firefly Video Model announced for late 2024 GA. Deepfake risks and content authenticity concerns documented by researchers. Market consolidation continued amid persistent production-readiness questions.
  • 2024-Q4: Pika 2.0 launched (Dec 13) with Scene Ingredients, reaching 11M users; Balenciaga, Fenty, Vogue deployed AI-generated ads. Adobe Firefly Video Model achieved GA in Creative Cloud. HeyGen survey (N=2,385) found 90.9% consumer comfort with branded AI videos. Cross-sector deployment metrics: 42% of advertisers report 60%+ cost reductions; Netflix cut trailer dev from weeks to hours; SAP reduced onboarding costs by 80%. Creator concerns (identity dilution, creative control loss) documented by CHI 2024 study; production limitations and practitioner skepticism remained unresolved.
  • 2025-Q1: Runway Gen-4 launched (March 31) with improved character/location consistency and visual reference integration. Pika 2.5 released with enhanced motion control and consistency for short-form workflows. Market paradox emerged: 60% of marketers use AI video daily but 95% fail to show ROI; independent tool testing found only 3 of 23 tools production-ready. Copyright litigation against Runway; film industry job displacement concerns escalated. Production-readiness barriers (sound, physics, consistency) remained unresolved despite adoption growth.
  • 2025-Q2: Pika 2.2 reached 11M+ users with enhanced keyframe transitions and extended clip length; Balenciaga and Vogue deployed for social ads. Google Veo 3 expanded to 71 countries with native audio. Market analysis showed Pika positioning for creators (500K Discord growth, 1B+ views), Runway for professionals (25M+ ARR). Critical assessments identified pricing friction (credit burn, account instability) limiting professional adoption. Practitioner testing (Civitai) confirmed persistent barriers: facial flickering, physics gaps, limited narrative control across multi-character scenes. Tool evaluation maintained 3-of-23 production-ready threshold, suggesting market hype outpacing technical maturity.
  • 2025-Q3: Runway Gen-4 Turbo deployed on WaveSpeedAI inference platform with 5x speed gains, demonstrating third-party ecosystem maturity. Pika 2.5 solidified creator positioning targeting influencer workflows ($100B+ creator economy). Technical assessments (July–Sept) reinforced persistent production barriers: temporal inconsistency, multi-character scene failures, synthetic audio gaps, and rapid credit burn (50% success rate, 187-second budget limitation) frustrating professional adoption despite capability advances. Market paradox persists: 60% daily AI video use but 95% corporate deployments lack sustained ROI; 3-of-23 tools production-ready. Governance barriers (copyright litigation, job displacement, content authenticity) remain unresolved.
  • 2025-Q4: Runway Gen-4.5 launched (Dec 1) achieving 1,247 Elo benchmark leadership with improved motion and prompt adherence. Pika Labs reached 14.5M users with sustained brand adoption (Balenciaga, Fenty, Vogue). Market adoption accelerated: 342% YoY growth, $4.2B market size, 80-95% cost reductions. Quality threshold crossed: 73% of viewers cannot distinguish AI from traditional video; AI-enhanced UGC achieves 45% engagement increases. Bifurcated maturity confirmed: cost-optimized marketing workflows production-ready, but temporal consistency, engagement trade-offs (15-30% lower), and governance risks (copyright, deepfakes) persist, constraining expansion to regulated and creative-critical use cases.
  • 2026-Jan: Runway Gen-4.5 integrated with Adobe Firefly, broadening deployment reach into mainstream creative workflows. Independent testing (CalMatters/The Markup) of four leading models reveals production-quality boundaries: none accurately generate complex choreography; 11 of 36 test videos exhibit consistency failures; cultural authenticity concerns documented. Ecosystem maturation signals: short-form video script market reached $2.74B (2026) at 29.9% CAGR; A-Roll/B-Roll production patterns standardized. Field testing confirms persistent quality variations and infrastructure friction (queue delays, credit burn issues) limiting professional adoption despite accelerating marketing use.
  • 2026-Feb: Adoption inflected dramatically: monthly AI video generation surged 5x (62K orders Jan 2026, 120K+ videos tracked). Google Veo 3.1 captured 96.4% market share via free bundled access; enterprise deployment reached 42% among Fortune 500 marketing departments. YouTube campaigns showed 17% ROAS improvement with AI optimization; creator economies accelerated (India $15B projected, 68% Gen Z weekly use). Production workflows standardized but infrastructure barriers persisted: Pika's credit system frustration, Adobe Firefly synthetic quality, 60% daily use but 95% ROI failures.
  • 2026-Mar/Apr: Runway Gen-4.5 reached critical maturity milestone: human perception study (n=1,043) showed 90% cannot distinguish AI from real video, with 57.1% detection accuracy. Real-world deployments validated: Animatic Media's AI-produced channel reached 9,900 monthly viewers; a 340-SKU fashion brand achieved 80% CVR lift and 99% cost reduction ($0.01–0.50/SKU vs. $500–1,500 traditional). Runway's $315M Series E (Feb 2026, $5.3B valuation, $860M total raised) underpinned named professional deployments — 20th Century Studios (Alien: Romulus VFX), Netflix Japan, Toei Animation, Bandai Namco, and Sony Pictures Japan — with production time savings from weeks to hours. Market scale confirmed: $18.6B AI video market (34.2% CAGR), 840% volume growth since Jan 2024, 73% of Fortune 500 using AI video tools. However, OpenAI Sora shutdown (March 24, 2026) exposed structural economics failure: $1M/day inference costs, user base declining from 1M to under 500K, and free-tier users generating exploratory content that never converts to paid workflows. YouTube's response to 1M+ channels producing AI content daily — prioritising a battle against "AI slop" — signals that platform quality enforcement is now a mainstream governance issue. Production costs stratified sharply: hybrid workflows fell 35-45% ($1.2-4.5K/min), AI-native fell 85-92% ($150-800/min), enabling 15-30 variant workflows per campaign. Enterprise B2B shifted toward human-led production with AI assistance, rejecting pure AI-generated video for trust-critical content. Academic surveys (HKUST/Tsinghua/Tencent) confirmed controllability remains the primary research frontier; content provenance standards and copyright litigation remain unresolved barriers.
  • 2026-May: Independent benchmarking (The Editorial, 240 prompts) placed Sora 2 and Veo 3.1 at the top for photorealistic text-to-video, with Runway Gen-4.5 and Kling 3.0 scoring 9.4/10 and 9.3/10 respectively — the first multi-model benchmark to treat these tools as production-comparable rather than experimental. IAB data confirmed U.S. digital video ad spend reached $80B (up 11% YoY) with two-thirds of video buyers activating AI, while field analysis of TikTok viral content showed Veo 3 leading UGC at 87% engagement parity; however, 67% of AI-generated content still underperforms human-produced equivalents, sustaining the quality stratification that has defined the practice throughout 2026.

TOOLS