AI Marketing OS

Three cycles, one surface: how the AI Marketing OS runs.

Theme-driven marketing, run as three cycles. Monthly, the themes are scored on the market and on meaningful traffic, and the programs under them are planned. Weekly, the content and the campaigns are made. Daily, the numbers are pulled, read and turned into insights that feed the next cycle. Marketing applications record what happened; the agents of the AI Marketing OS read across them, decide what happens next and act; a person approves the outcomes. That is the paradigm shift in one function, and we run it on our own marketing first.

The screens below are the surface itself: the content pane on the left two thirds, the conversation pane on the right third. Aldermere Supply Co. is an illustrative client; every number is mock data.

Once a month the agents read the market and re-score every theme; the programs for the month are planned under the themes that hold their score. A person approves the theme set, the slate and anything with its own date. Nothing is written to the plan before that.

58 Agent steps
3 Decisions
4 Programs
ai marketing os aldermere supply co.
SetupMarket map Proposed from the website
Aldermere Supply Co.Commercial kitchen fittings, distribution and light manufacturing
www.aldermeresupply.com412 pages read, 38 product families

Competitors

from the comparison pages

  • Brightline Fittings
  • Corvid Commercial
  • Hallam & Reed
  • Northgate Kitchen Systems
  • Pemberly Supply
  • Vantrell

Partners

the dealer networks the site names

  • Foodservice Equipment Dealers Alliance
  • Southern Kitchen Dealers
  • Pacific Restaurant Supply Group

Industry news

trade press the pages cite

  • Foodservice Equipment Monthly
  • Kitchen Trade Weekly
  • Restaurant Operations Journal
  • The Dealer Desk

Analysts

  • Hargrove Foodservice Research
  • Linden Market Advisors

Audiences

who the pages speak to

  • Dealer principals
  • Facility and operations managers
  • Restaurant group buyers

Two questions the site cannot answer

  1. Share of sales direct and through dealers
  2. Regions to grow this year
ThemesOctober review Awaiting reviewApproved, 4 themes committed
ThemeFamilyMarketTrafficScoreCoverageState
Lead time is the productOutcome786472up 46 piecesKeep
Three stock counts that disagreeOutcome745867new0 piecesAddAdded
The dealer portal nobody usesOutcome697170up 24 piecesKeep
Spec sheets buyers can findTechnology618069down 19 piecesKeep
Service calls that never become ordersOutcome664959down 63 piecesRetire
Three stock counts that disagreeAwaiting reviewApproved

The warehouse, the web store and the accounts each hold a stock figure, and a buyer is quoted from the wrong one. The job is one count; the number is days from order to invoice; the risk is a promise date missed.

Evidence, read this run

  • Searches for "inventory sync" terms up 23 percent in the category this quarter
  • Two competitors publish on it; neither names the three systems
  • The 3PL pages earn 18 percent of organic sessions and no lead form

Path

Leads to the integrated inventory, orders and accounts path.

ProgramsOctober 2026Lead time, quoted and kept Slate proposed, awaiting your 1Slate approved, 3 programs active
October 2026October media: 1,680 dollars, 3 programs6 Oct13 Oct20 Oct27 Oct
Lead time, quoted and keptlead-time-2026q4 640 ProposedActive
One count, three systemsone-count-2026q4 560 ProposedActive
The dealer portal, openeddealer-portal-2026q4 480 ProposedActive
Lead time, quoted and kept lead-time-2026q4Awaiting your 1Active, approved 7:58 a.m. Pacific
Theme
Lead time is the product
Window
6 October to 2 November
Media budget
640 dollars, 22 a day
Kill rule
no discovery request by week four

Pieces

  • Landing page
  • Blog post
  • Video
  • Email issue
  • Social set
  • Search campaign
ProgramsOctober 2026Launches and news One launch added
October 2026October media: 1,680 dollars, 3 programs1,980 dollars, 4 programs6 Oct13 Oct20 Oct27 Oct
Lead time, quoted and keptlead-time-2026q4 640 Active
One count, three systemsone-count-2026q4 560 Active
The dealer portal, openeddealer-portal-2026q4 480 Active
QuickFit rail launchquickfit-launch-2026-10 300 Added
QuickFit rail launch quickfit-launch-2026-10Added to the month
Date
Tuesday 20 October
Makes
a press release, a launch email, a landing page
Release
published 15 minutes before the email goes out
Media budget
300 dollars, two weeks

Pieces

  • Press release
  • Launch email
  • Landing page
  1. The market map

    The method starts with the market, read from the website.

    A company gives the surface its website and the agents read it: the industry, the product families, the competitors the pages compare against, the partners they name, the trade press and the analysts who cover the category, the audiences each page speaks to. A proposed market map comes back; the only questions asked are the ones the site cannot answer.

  2. Theme scoring

    Every theme is scored on the market and on meaningful traffic.

    A theme is a claim about what the market is wrestling with, and every piece under it is evidence for the claim. Each month the agents re-score the set on market pull (size, growth, switching, the installed base) and on meaningful traffic (the searches, pages and leads a theme already earns). Themes survive on their score, not on the calendar. The set is proposed; a person approves it.

  3. Programs

    Programs are the unit of work and the unit of measurement.

    A program is one argument under one theme, run across the channels for a fixed window: a landing page, a post, a video, an email issue, a social set and a search campaign. It is the smallest thing whose results are a real signal, and its slug is the join key every number carries. The slate lands on the calendar; a program opened shows its pieces, its budget and its kill rule.

  4. Launches and news

    Launches, releases and events join the month as programs too.

    Not everything is cyclical. Before the plan closes, the agents ask about product launches, press releases, events and anything else with its own date. A launch becomes a program with a release, a dedicated email and a landing page, placed on the calendar beside the themed work, so the month is one plan with one budget.

Every week the agents draft the pieces the active programs call for, one of each type per program, and build the campaigns that carry them in the applications that run them. A person reviews drafts and approves spend; the caps hold in code below the gate.

61 Agent steps
3 Decisions
4 Campaigns
ai marketing os aldermere supply co.
ContentWeek 41 slate 12 of 20 drafted, 0 approved
Pieces this week20one of each type per program
Drafted12ready for a reader
Drafting8by Wednesday
Approved0each on its own digit
Week 41PagePostVideoEmailSocial
Lead time, quoted and kept StagedReadyScriptDrafted8 drafted
One count, three systems StagedDraftedDraftingDraftedDrafting
The dealer portal, opened DraftingDraftedDraftingDraftingDrafting
QuickFit rail launch DraftedReleaseNoneDraftedNone
StatesDraftingDrafted or readyApproved or staged
ContentLead time, quoted and keptBlog post In reviewApproved, scheduled Tuesday 7:00
Blog postlead-time-2026q4Lead time is the product

Why your lead time is the product your buyers compare.

Dealers do not compare fittings. They compare the date they can promise a kitchen, and whether it holds.

Cover: a quoted date on a dealer order, kept.

Every quote carries a date. The buyer reads the date before the price, because a kitchen that opens two weeks late costs more than any fitting in it. The lead time is the product; the catalogue is how it is delivered.

Three things decide whether a quoted date holds: the stock count the quote was read from, the carrier window the 3PL can keep, and whether the order and the invoice agree. Each lives in a different application, and the gaps between them are where the date is lost.

CampaignsWeek 41Spend Spend awaiting your 1Enabled, 69 dollars a day
AdsBuilt, pausedEnabled
lead-time-2026q4
Daily budget
22 dollars
Ad groups
3, 48 keywords
Lands on
/lead-time
Marketing automationBuilt
Lead time journey
Emails
3 over 14 days
Audience
1,840 dealer contacts, CRM
Starts
with the post, Tuesday
SocialScheduled on approvalScheduled
Week 41 schedule
Posts
8, Monday to Friday
Window
7 a.m. to 4 p.m. Pacific
Carry
the post, the video, the page
SiteStagedLive
/lead-time
Landing page
staged, 1,120 words
Tag
utm_campaign lead-time-2026q4
Form
discovery request
Week 41 spendAwaiting your 1Approved 7:58 a.m. Pacific
Search campaigns
4, one per program
A day
69 dollars: 22, 20, 17 and 10
The week
483 dollars
Cap in code
100 dollars a day for the account
  1. Content

    One of each piece per program, every draft to its own gate.

    For every active program the week holds one blog post, one video, one email issue, one social set and one landing page, drafted by the agents from the program’s argument and the theme’s evidence. The slate shows each piece in its state; a draft is reviewed where it was made, and nothing publishes without a digit on that piece.

  2. The draft

    Drafts open where the work is, never in another application.

    A post, a script, an email issue, a social set or a landing page opens inside the surface: the copy, the cover, the search fields and the schedule, with the program and the theme it serves. The review happens here with the conversation beside it: approve, decline or say what to change, and the agent revises in place.

  3. Campaigns

    Every campaign through the one surface, every dollar behind a gate.

    Search campaigns, marketing automation journeys, the social schedule and the landing pages are built by the agents in the applications that run them and shown as panes of the one surface, never as a vendor’s own screen. Spend has a cap enforced in code below the gate; the week’s spend is asked for once, with the numbers, and acted on from the digit.

Every morning the agents pull the numbers from every source into one table, read it and write what they see, with a diagnosis. Reports open in place; an insight that calls for a change comes back as a proposal with its numbers and waits for a digit. What is learned returns to the week and the month.

31 Agent steps
1 Decisions
5 Sources
ai marketing os aldermere supply co.
ResultsThursday 16 OctoberObservations Pulled 7:10, 3 observations
The daily pullThursday 16 October, 7:10 a.m. Pacific
  • Analytics96 rows, 4 programsRead
  • Ads48 rows, 4 campaignsRead
  • CRM14 leads, 4 sales-readyRead
  • Email2 sends, 3 journeysRead
  • Social52 posts taggedRead
  1. 1

    Search cost per lead on The dealer portal, opened rose to 131 dollars from 96 last week.

    Diagnosis. The kitchen-rail ad group took 58 percent of the program’s spend for one lead; its broad match is pulling consumer searches.

    Open the ad group, day 4
  2. 2

    Lead time, quoted and kept: 1,240 engaged sessions, up 31 percent on last week; two sales-ready leads.

    Diagnosis. The post and the first email issue landed in the same 48 hours; the journey is doing its job.

    Open the program report
  3. 3

    Email issue 1 of Lead time: 38 percent opened, 6.1 percent clicked.

    Diagnosis. Southeast dealers clicked at twice the average, the region named in the setup.

    Open the send, by region
ResultsLead time, quoted and keptProgram report Day 11 of 28, on track
Lead time, quoted and keptlead-time-2026q4, 6 October to 2 November, day 11 of 28
Engaged sessions1,240up 31 percent on last week
Leads64 from search, 2 from email
Sales-ready2one discovery request
Cost per lead40dollars, media only
Media spent242of 640 dollars
All-in cost611media plus the running cost of the pieces
  • Blog postWhy your lead time is the product your buyers comparepublished 7 Oct, 412 readsOpen the post
  • VideoThe date before the pricepublished 9 Oct, 318 viewsOpen the video
  • Email issueIssue 1, the quoted datesent 8 Oct, 38 percent openedOpen the send
  • Social set8 posts2,140 impressionsOpen the schedule
  • Landing page/lead-time1,240 engaged sessionsOpen the page
  • Search campaignlead-time-2026q4242 dollars, 4 leadsOpen the campaign
Kill ruleno discovery request by week four: one request on day 9
ResultsThe dealer portal, openedProposed change Proposal awaiting your 1Done, read back 7:31
Move 8 dollars a day to facility managersAwaiting your 1Done, read back from the campaign 7:31

Observation

Search cost per lead on The dealer portal, opened is 131 dollars against 96 last week. The kitchen-rail ad group took 58 percent of spend for one lead; broad match on rail terms is pulling consumer searches.

Proposed change

Move 8 dollars a day from the kitchen-rail ad group to the facility-manager ad group, and tighten the rail terms to phrase match. The program’s 17 dollars a day and the account cap are unchanged.

Before

Kitchen rail14dollars a day, 1 lead
Facility managers3dollars a day, 3 leads
Cost per lead131dollars

After, expected

Kitchen rail6dollars a day, phrase match
Facility managers11dollars a day
Cost per lead95dollars, expected
Cap in code. 100 dollars a day for the account; the change stays under it The observationThe ad groupThe campaignThe program report Learning kept. consumer searches on rail terms; added to the theme’s evidence for the November review
  1. The pull

    One table, every source, every morning.

    Analytics, search, the CRM, email and social are read once a day into one performance table, one row per day per program, joined by the program’s slug. Then the agents read the table and write what they see: not a dashboard to interpret but observations with a diagnosis, each pointing at the row behind it.

  2. Reports

    Reports and the files behind them open in place.

    A program report is a page of the surface, not an export: engaged sessions, leads, sales-ready leads, cost per lead and all-in cost, against the program’s window and kill rule. The pieces and campaigns it cites are links that open here, the post, the campaign, the journey, so a number is never more than one click from the thing that made it.

  3. Insights

    An insight becomes a proposed change, with its numbers, waiting for a digit.

    When an observation calls for action the agent writes the change it would make and what it expects: move the budget, pause the ad group, retitle the page, resend to a segment. The proposal shows the before and the after. On a 1 it is done and read back; on a 3 it is changed first. Every answer is logged, and the learning returns to the slate and the themes. Applications record; agents act; a person approves outcomes, not hours.

The operating model

What a person decides, and what the agents do.

Outcomes, not hours. The cycles leave a person with the decisions and the agents with the work between them, every step in one log.

A person decides.

  • The theme set, once a month
  • The slate and anything with its own date
  • Each draft, on the piece
  • The spend, once a week, under a cap in code
  • Each proposed change, with its numbers

The agents do the rest.

  • Read the market and re-score the themes
  • Plan the programs and the calendar
  • Draft every piece and build every campaign
  • Pull the numbers into one table every morning
  • Observe, diagnose, propose and read back
  • Log every step with who approved it

Marketing applications record what happened. The agents of the AI Marketing OS read across them, decide what happens next and act, and a person approves the outcomes. It is the paradigm shift in one function; we built it for our own marketing first, and we run it for clients as a managed solution.

Aldermere Supply Co. is an illustrative client. Every figure on the screens is mock data.

How we start

Most of our best clients come to us with a feeling, not a plan.

"Something isn't working." "We're outgrowing our tools." "We're afraid to make the wrong move." No-Risk Discovery is a short, practical conversation that gets you clarity before you commit to anything big. We'll tell you if we're a fit. If we're not, we'll tell you that too.