Agentic Maturity Index · rebuilt weekly · dataset v2026-11-15.30 · method 1.1

Who is moving fastest.

Every company in the Atlas with an agent interface, scored 0–100 on what those interfaces document an AI agent can do. A classic API's permissions do not count toward the agent score. Vendor momentum needs a dated launch with a primary source; Atlas research and corrections are tracked separately. This measures documentation, not production performance.

41.2/100 ▲ 1.0 vs 90 days ago, reconstructed
  • 158scored: an operational agent interface
  • 16listed, not ranked (12 classic API only, 4 reference only)
  • 30dated agent launches, last 90 days
  • MGIDfastest mover

Edition mx-1.1-2026-11-15 published 15 Nov 2026 · data as of 15 Nov 2026 · newest source re-check 15 Nov 2026 · source watch last read 373 cited pages on 13 Nov 2026. A weekly rebuild is not a re-check of every claim.

Reconstructed index by month, with dated agent-interface launches3740432025-06: 1 dated agent launches2025-07: 2 dated agent launches2025-08: 1 dated agent launches2025-09: 2 dated agent launches2025-10: 1 dated agent launches2025-11: 1 dated agent launches2025-12: 2 dated agent launches2026-01: 2 dated agent launches2026-02: 1 dated agent launches2026-03: 2 dated agent launches2026-04: 10 dated agent launches2026-05: 6 dated agent launches2026-06: 19 dated agent launches2026-07: 5 dated agent launches2026-08: 4 dated agent launches2026-09: 12 dated agent launches2026-10: 13 dated agent launches2026-11: 4 dated agent launchesbars: 19 launchesJun 25Sept 25Dec 25Mar 26Jun 26Sept 26Nov 26
Reconstructed index at month end Dated agent launches that month

A reconstruction, not a record: today's documented facts placed at each interface's launch date, with the 113 undated agent interfaces assumed present throughout. Published weekly values start in November 2026.

Vendor momentum · last 90 days

Fastest movers

Points gained from interfaces with a dated launch in the last 90 days. Undated interfaces count at both ends, so only real, dated moves show.

  1. ▲ 68
    MGID

    0 → 68

    • MGID Advertiser API MCP · 10 Nov 2026
  2. ▲ 68
    MNTN

    0 → 68

    • MNTN MCP · Oct 2026
  3. ▲ 60
    Yahoo DSP

    0 → 60

    • Yahoo DSP MCP Server · 23 Oct 2026
  4. ▲ 48
    Adzymic

    0 → 48

    • AgenX Sales Agent · 20 Oct 2026
  5. ▲ 46
    Sovrn

    0 → 46

    • Curation MCP · 25 Sept 2026
  6. ▲ 45
    Draft Digital

    0 → 45

    • Own buying agent used with Mediahuis · 18 Sept 2026
  7. ▲ 44
    Mediahuis

    0 → 44

    • Seller agent used with Draft Digital · 18 Sept 2026
  8. ▲ 38
    Everflow

    0 → 38

    • Everflow Network MCP Server · 10 Nov 2026

Leaderboard · rank change since 14 Nov 2026

The index, company by company

Showing the top 25 of 158 scored companies

  • Agent reach /20
  • Capability depth /40
  • Task authority /25
  • Access /10
  • Evidence /5
  1. =1 – AdaptyMobile app growth 80 —
  2. =1 – SkaiCampaign management 80 ▲ 34
  3. =3 – TaboolaDSP 78 —
  4. =3 – TikTokWalled gardens 78 —
  5. =3 – VibeDSP 78 —
  6. =3 – XWalled gardens 78 —
  7. 7 – ApostraAgent-native 77 —
  8. 8 – MagniteSSP & curation 73 —
  9. 9 – MetaWalled gardens 72 —
  10. 10 – EquativSSP & curation 71 —
  11. 11 – PropellerAdsDSP 70 —
  12. 12 – Amazon AdsWalled gardens 69 —
  13. =13 – FreeWheelSSP & curation 68 —
  14. =13 – Google Ad ManagerSSP & curation 68 —
  15. =13 ▲2 MGIDDSP 68 ▲ 68
  16. =13 ▲14 MNTNDSP 68 ▲ 68
  17. =17 ▼2 AdspirerCampaign management 66 —
  18. =17 ▼2 SupermetricsCampaign management 66 —
  19. =17 ▼2 Windsor.aiCampaign management 66 —
  20. =17 ▼2 XnurtaCampaign management 66 —
  21. =21 ▲33 MadhiveDSP 65 —
  22. =21 ▼1 PubMaticSSP & curation 65 ▲ 26
  23. =21 ▼1 SuperwallMobile app growth 65 —
  24. 24 ▼2 IntentwiseCampaign management 62 —
  25. =25 ▼3 InfillionDSP 60 —

Rank markers compare with the snapshot of 14 Nov 2026, one day earlier; a rank can move with an unchanged score. 90-day chips are vendor momentum. Sparklines are reconstructed month-end scores on a shared month axis.

Listed, not ranked: 16 companies

Classic API only (12). Operational APIs or SDKs but no operational agent interface yet, so there is nothing for the agent score to measure: DoorDash, Etsy, Index Exchange, Innovid, Instacart Ads, JWX, OpenAI, OpenX, Overwolf, Uber, Viant, Walmart Connect.

Classification queue (81). Agent records that carry no explicit interface class yet. They are scored as operational and listed for review; a missing class is never read as invalid or as certified.

Reference only (4). Only documentation, listings or adjacent interfaces in the Atlas: atomicAds, HERE Technologies, Klarna, OneTrust.

By category

Where the market is mature, and where it is not

CategoryAverageAgent reachCapability depthTask authorityAccessEvidence
Agent-native 4 56 75%53%38%70%55%
Campaign management 14 51 67%46%33%71%70%
Walled gardens 11 51 62%52%26%64%89%
DSP 17 50 62%49%36%55%80%
Commerce platforms 4 48 85%50%0%80%60%
Mobile app growth 11 44 64%47%9%73%60%
Data collaboration & identity 4 44 65%38%24%60%80%
Customer engagement & activation 9 42 67%46%6%58%64%
Cloud, data & AI platforms 4 39 65%38%0%80%60%
SSP & curation 26 39 66%34%20%32%74%
Location & geospatial 7 34 71%29%0%54%60%
Measurement & audience data 35 33 63%30%4%44%62%
Creative production & ad serving 5 30 60%30%0%32%60%
Retail media 7 30 60%23%0%51%66%

The agentic wave

Dated agent launches, month by month

MCP servers and AdCP, AAMP or A2A agents with a vendor-dated launch: 88 of 201 operational agent interfaces carry a date (2 to the month only, counted from the last day of that month). The other 113 count in the index but cannot be placed in time, so this chart is not a record of when the market grew.

2025-06: 11Jun 252025-07: 222025-08: 11Aug 252025-09: 222025-10: 11Oct 252025-11: 112025-12: 22Dec 252026-01: 222026-02: 11Feb 262026-03: 222026-04: 1010Apr 262026-05: 662026-06: 1919Jun 262026-07: 552026-08: 44Aug 262026-09: 12122026-10: 13132026-11: 44Nov 26

Since the previous snapshot · one day

Score changes since 14 Nov 2026

14 Nov 2026, index 40.7 (158 scored) → 15 Nov 2026, index 41.2 (158 scored), both under method 1.1. A score change can come from a vendor launch, new Atlas research or a correction; the change feed says which. 87 companies moved rank with an unchanged score because others changed or entered; they are not listed here.

Index +0.5 = +0.4 from score changes at the 158 companies scored in both snapshots +0.1 from 0 entering, 0 leaving and roundingThe 158 companies in both moved +0.4 on average. Snapshot mx-1.1-2026-11-15, observed, input 9b5c0c5a0ee23e31.

Score up

Score down

  • No score decreases in this period.

New to the index

  • No new companies in this period.

Every change, with its source, in the change feed

How the index is built

Five components, agent interfaces only

  • Agent reach · 20Operational MCP, AdCP, AAMP or A2A interfaces: 12 for the first, 16 for two, 20 for three or more.
  • Capability depth · 40Best settled state across those agent interfaces for read, write, optimize and activate: documented 10, limited 5. Not established, excluded or unresolved earns 0.
  • Task authority · 25An agent interface itself qualifies for: change a campaign 10, change bids, budgets or pacing 8, put an audience live 7. A classic API does not lend its permissions.
  • Access · 10Best access tier of an agent interface: open 10, any customer account 8, account plus enablement 6, partners or by request 4, invite or staged rollout 2, not established 0.
  • Evidence · 5Share of agent interfaces re-checked against sources (up to 3), plus 2 when one is reviewed capability by capability with no positive capability check unresolved.

A company's score uses only its operational agent interfaces (MCP, AdCP, AAMP, A2A). Every point comes from the same interface that documents it: a broad classic API does not lend its permissions to a read-only MCP. Companies with only classic APIs, and reference-only records, stay in the Atlas but are listed, not ranked, and the index averages the 158 scored companies.

Missing evidence earns no points. A capability the documentation does not establish scores zero until a primary source settles it; that is not a finding that the capability is absent. A positive capability claim whose check is unresolved counts as unresolved, not as support, in capability depth and task authority. Access uses the recorded access tier, which carries its own evidence status; an unresolved access, sign-in or date check is shown in the company's Why panel, not hidden. Vendor momentum compares today with 90 days ago using dated launches only.

Method 1.1 (15 Nov 2026): Agent interfaces only; unsettled claims count as unresolved; reference-only and classic-API-only companies are listed, not ranked. Earlier snapshots were restated under it from the Atlas's git history and are marked as restated; the method 1.0 values are kept unchanged in the history archive. Revisions only compare snapshots scored with the same method, so a method change never shows up as a vendor change. Each published snapshot is frozen with its input hash and commit; a later finding produces a new snapshot or a labelled restatement, never an edit. Every change in the index is split into score changes at companies present in both snapshots and the effect of companies entering or leaving. A launch dated only to the month counts from the month's last day. It is a measure of what is documented, not a certification of what works in production.

Request the full dataset Sample data (JSON) Atlas method Explore the Atlas

Full dataset

Every company, every component, every week

The public JSON is a sample: the headline index, its weekly series, category averages and the top ten. The full dataset has every scored company with its five component scores, 90-day momentum with the dated launches behind it, and each company's score and rank in every weekly snapshot, as JSON and CSV. Tell us who it is for and Evgeny will reply with the files and terms of use.

Ask for it through the contact form: say who it is for and what you will use it for.

Cite as: No Fluff Advisory, Agentic Maturity Index, 15 Nov 2026, dataset v2026-11-15.30, https://nofluffadvisory.com/agentic-atlas/maturity/. CC BY 4.0.