Business AI Market Observatory

Feature diffusion

Which capabilities spread across the panel, how fast, and who moved first. Features are detected from pricing-page text with a versioned lexicon; adoption dates are intervals between consecutive snapshots.

Adoption share by quarter

Share of products observed in a quarter whose pricing page matched the feature. Sample size per quarter shown on hover.

feature (% of products)2022Q12022Q22022Q32022Q42023Q12023Q22023Q32023Q42024Q12024Q22024Q32024Q42025Q12025Q22025Q32025Q42026Q12026Q22026Q3
Knowledge retrieval (RAG)47454441414043384040394142464439403847
AI agents00022225710152027344140434449
Voice interaction20202021222627242423252223292931303234
Image input / vision4663101513141415161819191817161716
Browser / computer control2222232100122335686
Coding agent00000221000237910111419
Persistent memory989121012998771011111314151516
SSO / SAML33333633323737364039444244464745515247
SCIM provisioning7899897666699131614171821
Audit logs7891010141512911111616172018263029
Data residency0000000022222342488
Compliance certifications27373836372925282931313433383735374340
No-training / zero data retention00000031337779978911
Admin console20181614141213161918192222232728333336
API access20222624242321252627263030313028292927
Custom / fine-tuned models9141412131413151622232426252419201612
Multiple model choice00002023124778865610
Web search / browsing0002200323765889789
Deep research00002201113466767910
Image generation22798813101011131315191919192023
Video generation4443356554579111314141517
Workflow automation27283138413935383439425048505251565959
Third-party integrations64637369767166636563636567716665646973
MCP support00000000000012511122025
Canvas / artifacts96653334333337978910
Credit-based usage49495148525253464747475255505254545960
Reasoning models000000000014710148979
Knowledge retrieval (RAG)
AI agents
Voice interaction
Image input / vision
Browser / computer control
Coding agent
Persistent memory
SSO / SAML
SCIM provisioning
Audit logs
Data residency
Compliance certifications
No-training / zero data retention
Admin console
API access
Custom / fine-tuned models
Multiple model choice
Web search / browsing
Deep research
Image generation
Video generation
Workflow automation
Third-party integrations
MCP support
Canvas / artifacts
Credit-based usage
Reasoning models
Quarterly product counts: 2022Q1: 45 · 2022Q2: 51 · 2022Q3: 55 · 2022Q4: 58 · 2023Q1: 63 · 2023Q2: 65 · 2023Q3: 68 · 2023Q4: 81 · 2024Q1: 89 · 2024Q2: 94 · 2024Q3: 100 · 2024Q4: 103 · 2025Q1: 108 · 2025Q2: 108 · 2025Q3: 110 · 2025Q4: 109 · 2026Q1: 108 · 2026Q2: 106 · 2026Q3: 115. Early quarters rest on a handful of products.

Convergence (RQ1)

Mean pairwise Jaccard similarity of tracked feature sets, all observed products vs the balanced sub-panel observed in every recent quarter.

All observed products (95% bootstrap CI)
Balanced panel (91 products in every quarter 2024Q42026Q3)
Mean number of tracked features per product (± 1 SD)
Linear trend of the quarterly mean Jaccard: +0.0004 per quarter (SE 0.0005, t = 0.69, 19 quarters). Descriptive; composition changes over time.

Time to adoption (Kaplan–Meier)

Months from a product's first pricing snapshot to first detection of the feature. Products that already showed the feature at first observation are left-censored and excluded from the curve (counted separately); never-adopters are right-censored at their last snapshot.

Eight features with the most in-panel adoption events; all features are in the table below
FeaturePrevalence now (filtered)AdoptersAlready present at first obs.Adopted inside panelFirst adopters (by upper date)Median lag from first (months)Within 6 / 12 months of firstKM at risk / eventsKM median (months)
Third-party integrations90 / 1251076146zendesk-ai (2022-01-01†), google-workspace-gemini (2022-01-01†), canva (2022-01-02†)333 / 961 / 4617.9
Workflow automation66 / 125893455zendesk-ai (2022-01-01†), canva (2022-01-02†), synthesia (2022-01-04†)32.42 / 888 / 5526.8
Credit-based usage66 / 125864838zendesk-ai (2022-01-01†), google-workspace-gemini (2022-01-01†), grammarly (2022-01-02†)35.81 / 474 / 3837.5
Knowledge retrieval (RAG)55 / 125884345zendesk-ai (2022-01-01†), google-workspace-gemini (2022-01-01†), tidio-lyro (2022-01-04†)333 / 779 / 4532.5
AI agents50 / 12570862zendesk-ai (2022-11-01), beam-ai (2023-10-03†), relevance-ai (2023-12-06)301 / 1114 / 6242.8
SSO / SAML48 / 125732845zendesk-ai (2022-01-01†), grammarly (2022-01-02†), canva (2022-01-02†)32.41 / 394 / 4541
Compliance certifications47 / 125652837zendesk-ai (2022-01-01†), grammarly (2022-01-02†), deepgram (2022-01-05†)29.15 / 794 / 3752.6
Voice interaction35 / 125511833zendesk-ai (2022-01-01†), google-workspace-gemini (2022-01-01†), grammarly (2022-01-02†)39.90 / 1104 / 33
API access33 / 125532231zendesk-ai (2022-01-01†), synthesia (2022-01-04†), deepgram (2022-01-05†)291 / 5100 / 31
Admin console31 / 125541539google-workspace-gemini (2022-01-01†), otter (2022-01-10†), figma (2022-01-11†)38.60 / 1107 / 39
Audit logs29 / 12543538zendesk-ai (2022-01-01†), figma (2022-01-11†), writer (2022-01-17†)43.22 / 2117 / 38
MCP support25 / 12533132windsurf (2025-03-01), zapier (2025-05-04), augment-code (2025-05-05)13.66 / 14121 / 32
Custom / fine-tuned models23 / 125461630zendesk-ai (2022-01-01†), openai-api (2022-01-11†), hugging-face (2022-01-17†)27.61 / 3106 / 30
Image generation23 / 12538830openai-api (2022-01-11†), replicate (2022-07-31†), assemblyai (2022-08-20)37.80 / 3114 / 30
Image input / vision19 / 12542834openai-api (2022-01-11†), labelbox (2022-01-24†), lattice (2022-05-12)30.52 / 4114 / 34
Video generation18 / 12524420synthesia (2022-01-04†), runway (2022-02-12), heygen (2023-03-16†)40.21 / 1118 / 20
SCIM provisioning17 / 12534628figma (2022-01-11†), notion-ai (2022-01-11†), lattice (2022-01-31†)40.60 / 1116 / 28
Persistent memory16 / 12530624wandb (2022-01-19†), deepnote (2022-01-27†), replit (2022-02-07†)40.20 / 2116 / 24
Coding agent15 / 12528127mosaicml (2023-05-07†), github-copilot (2024-11-01), windsurf (2024-11-13)33.20 / 0121 / 27
No-training / zero data retention13 / 12519217github-copilot (2023-08-01), chatgpt (2024-01-10†), tabnine (2024-05-09)24.11 / 3120 / 17
Multiple model choice10 / 12524222deepgram (2023-02-28), perplexity (2023-09-01), scispace (2023-10-27)24.11 / 4120 / 22
Deep research10 / 12520119alphasense (2023-01-29), clay (2024-07-18), mistral-platform (2024-08-02)291 / 1121 / 19
Canvas / artifacts9 / 12519613canva (2022-01-02†), figma (2022-01-11†), notion-ai (2022-01-11†)42.10 / 0116 / 13
Browser / computer control8 / 12514212tidio-lyro (2022-01-04†), zendesk-ai (2023-05-05), clay (2024-07-18)46.50 / 0120 / 12
Web search / browsing7 / 12518018jasper (2022-12-03), github-copilot (2024-03-01), chatgpt (2024-06-01)29.41 / 1122 / 18
Data residency6 / 12513013assemblyai (2024-02-07), akkio (2024-02-22), chatgpt (2025-03-04)24.92 / 2122 / 13
Reasoning models6 / 12519118cody (2024-09-17), cursor (2024-10-01), openai-api (2024-10-01)7.48 / 15121 / 18
† = already present at that product's first snapshot (left-censored), so "first adopter" is really "earliest observed". Imitation lags are measured from the earliest observation in this panel, which starts in 2022 for most products; features introduced before then cannot be dated.