Status: Original concept, first publication 18th May 2026, updated with Kalicube Summit 2026 interview material 29th May 2026.

The richest source of evidence the AI uses to recommend your brand isn’t your blog, isn’t your structured data, and isn’t your topical hub. It’s everything your business naturally generates while running its operations: the onboarding calls, the performance numbers, the support resolutions, the unprompted client advocacy. Most of that material dies inside CRMs, support platforms, and quarterly retrospectives, and the few brands that capture it usually pipe it into testimonials nobody reads.

For me, this is the single biggest expansion of the SEO function in fifteen years, and almost nobody is running it as a structural practice. The signals AI engines weight most heavily when deciding whether to recommend you, deploy you, or reselect you are post-sale signals: onboarding accuracy, performance outcomes, integration depth, client devotion. None of that lives in your blog. All of it lives inside the operational core of your business, and most of it dies there.

The five stages where this material gets generated and packaged have a name. OPIDC. The first four describe what your business naturally does in the course of running its operations: Onboarded, Performed, Integrated, Devoted. The fifth, Codified, is what the SEO function does with what the business produces. This article covers all five.

OPIDC is the moment marketing, SEO, and business operations converge

OPIDC is articulated at the moment three disciplines that used to run in separate departments stop having the option to operate separately. The convergence has been building for a decade, and the dating chain is publicly retrievable. Answer Engine Optimisation (AEO), coined in 2017, was the moment SEO expanded into marketing, because the engine started surfacing a single synthesised answer that had to carry brand positioning rather than ten ranked links the user discriminated between. AI Engine Optimisation (AIEO), coined in 2024, pushed SEO further into product and operations, because the Intelligence layer reasoned over product positioning and the brand’s articulated value rather than over keyword-document matching. Assistive Agent Optimisation (AAO), coined in 2025, is where the convergence becomes mandatory: when an agent acts as the buyer, how the machine finds the brand, decides the brand is a credible resource, and physically transacts with the brand are technically the same question, and the discipline that runs each previously-separate step has to engage directly with pricing architecture, qualification gates, fulfilment mechanisms, and the commercial integration with agent-payment rails.

In every earlier era the convergence was available, not required. A business in the AIEO era could still run SEO, marketing, product, and operations in silos and survive, because the human user was still the discriminating audience and the silos could still hand off to each other through human-mediated friction. In the AAO era the silos cost real money, because the agent reads the brand, the marketing, and the machine-readable signal as one object and recommends or rejects on all three at once. Business operations can no longer ignore SEO and marketing either: the agent’s evaluation logic, optimised for fit to user request, treats operationally-illegible brands as illegible full stop, and the operational architecture (pricing, qualification, scheduling, fulfilment) becomes the substrate the agent is reading.

I’m articulating OPIDC in May 2026 because this is the moment the convergence has to be built, not the moment the convergence becomes mainstream. Agentic AI isn’t dominant yet. It will be soon, and the brands that have already converged their SEO, marketing, and business operations into one practice arrive at mainstream agentic with the discipline already running. The brands that wait arrive without it, and the catch-up is structural, not tactical.

The full articulation of this convergence, including the academic statement of how the disciplines stack (AAO ⊃ AIEO ⊃ AEO ⊃ SEO) and the strategic context that produces the convergence requirement, belongs in a separate Sandbox piece on AI-era Business Engineering. This article is about OPIDC, which is one of the operational mechanisms through which the convergence happens. Read OPIDC as the practice that makes the convergence actionable inside the business, not as the full theoretical case for why the convergence is now mandatory.

Search, Assistive, and Agential will cohabit indefinitely

A clarification before going further, because the convergence argument is sometimes misread as “everything becomes agentic.” It does not. Search mode, Assistive mode, and Agential mode will live in parallel, and most buyers will use all three across the course of a single purchase journey, sometimes within the same hour. A buyer researches a category in Assistive mode, navigates to a known site in Search mode, instructs an agent to replenish an existing supply in Agential mode. The same buyer in the same week occupies all three modes for different purchases. The shares will shift over time, agents will take more of the high-volume routine work, and the Agential share will keep growing as more categories support delegated decision-making, but the three modes are not a temporal sequence. They are co-present, and as far as I can see they will keep cohabiting indefinitely.

The structural concept that names where each buyer and each purchase sits in this mix is the delegation boundary: the line up to which a customer hands a decision to an agent. The delegation boundary sits in a very different place for a coffee shop (perhaps five per cent of custom will ever route through an agent) than for a SaaS platform delivering data (where it may eventually reach ninety-five per cent). The move toward that boundary is gradual, and where it settles depends on your geography, your type of service, and how urgent your buyers are. The practical question that sets your pace: to what extent has your ideal customer already delegated the choice of your product or service to an agent?

The implication for the substrate work is direct. A brand that engineers only for the agent loses the human-mediated encounters where the buyer wants to verify before authorising. A brand that engineers only for the human loses the agent-mediated encounters where the buyer never sees the candidate list the agent assembled. Each OPIDC stage has to serve both audiences (the human and the machine) because both audiences are present in the brand’s mix of encounters, and the proportion varies by product, by segment, by region, and by buyer.

OPID is the business, not a content opportunity

The four OPID stages are not passive stages the brand observes happening. They are the active core of business operation, and they’re where the business actually makes money. Onboarded is the operational practice of getting new clients into delivery successfully. Performed is the operational practice of delivering measurable outcomes against a baseline. Integrated is the operational practice of becoming structurally embedded in how the client works. Devoted is the operational practice of earning unprompted advocacy.

That’s the business. Sales doesn’t run those stages. Marketing doesn’t run those stages. Service, support, customer success, account management, delivery teams: they run those stages, every working day, and they are the operators who produce the material that decides whether AI engines recommend the brand to the next prospect.

What’s new is that the SEO function now has work to do INSIDE that operational core. Not running it. Harvesting from it. The four operational stages produce evidence as a natural byproduct. The SEO function’s job in the AI era is to make sure that evidence reaches the inference layer where AI engines read it, rather than dying inside the CRM where nobody reads it.

That distinction is load-bearing. Operators who treat OPID as “stages clients pass through” miss that OPID IS the business operation that funds the SEO function. Operators who treat Codified as “another content task” miss that Codified is the SEO function’s response to evidence the business produces structurally.

That framing also decides whether your service team will work with you. Walk into a customer-success meeting saying “I need content for my blog,” they’ll politely decline. Walk in saying “the evidence your team produces every week decides whether the AI recommends us to the next prospect, and I want to help you capture it,” they’ll lean in.

When OPIDC runs properly, you see the shape of it in unexpected places. James Dooley told me his sales team is now almost just filling in onboarding forms, because the AI has done the selling before anyone gets on a call. Inquiries down, sales up, salespeople bored because the buyer arrived already pre-sold. That’s the shape of OPIDC harvested and codified properly: the evidence does the work upstream, the AI carries it forward, and the human conversation just confirms what the buyer already decided.

One structural feature to hold across every stage: AI is now the second audience at every gate. In agential mode, the AI agent is the customer who arrives at Onboarded, evaluates Performed, observes Integrated, and decides Devoted before the human end-buyer has formed a clear opinion. Even when the buyer is human, AI is watching each stage and feeding the observation back into its training signal for the next prospect’s recommendation. Each OPIDC stage now has two simultaneous customers, the human and the machine, and the brand’s job is to satisfy both.

Onboarded sets the Satisfaction Gap

Onboarded is where the client starts. The onboarding team sets expectations, matches first-interaction success, and decides the gap between what marketing promised when the client bought and what the client actually experiences when delivery starts. That gap has a name in the framework: the Satisfaction Gap. Success here looks like its absence: first outcomes that match the sales narrative, early wins that arrive on time, specific quotes from the onboarding phase that prove the gap closed cleanly.

The material this stage produces is top-of-funnel content. It answers the question “what happens when I start working with X” with the specific evidence of what actually happens, in the client’s words, with verifiable outcomes. Brands that codify this material answer the prospect’s first AI query. Brands that don’t have AI engines hedging on their behalf.

Scott Duffy made a point in our Summit interview that reframes the whole stage: the work that decides whether Onboarded succeeds happens pre-sale, not post-sale. Two questions, asked before the contract is signed, define everything that follows. First, what’s most important to you about this engagement? Second, and the one almost nobody asks, what has to happen, how will you know you’re getting what you want? Without the second question, every team measures success differently from the customer, and Onboarded breaks in week three because everyone is running against a different scorecard. The fix is the buyer’s blueprint, captured pre-sale and held by every team that will touch the account. When the first internal report goes out, it doesn’t ask “how are we doing” in the abstract, it checks every box the customer wrote down before the engagement started.

Move: Ask your onboarding team to log one onboarding success per month with a specific client quote, the timeline from sale to first outcome, and what the client said the first outcome actually delivered against the buyer’s blueprint. Fifteen minutes per month from the onboarding team, one piece of bottom-of-funnel evidence per month walks straight into your codification pipeline.

Performed produces the numbers that survive a CFO’s scrutiny

Performed is where the service delivers. The delivery team produces real outcomes with real numbers: percentage improvements, time saved, problems resolved, milestones met, against documented baselines. These outcomes are middle-of-funnel gold because they answer the comparison question with evidence rather than claims. “Reduced support tickets by 43% in six months against a baseline of 1,200/month” is machine-legible evidence. “We helped them grow” is a claim the AI will hedge on forever.

If the delivery team isn’t tracking outcomes against baselines, the SEO function’s job is to make sure they start. Without numbers, there’s nothing to codify. The Performed stage produces the substrate for everything AI engines do at the consideration stage of the buyer journey.

Move: Sit with the delivery team. Identify which outcomes are already measured against baselines, which outcomes should be measured but aren’t, and what the smallest tracking change looks like to capture the missing baselines. The biggest blocker is usually that the team measures outcomes the client cares about (project completion, deliverable acceptance) but doesn’t capture the comparison against the client’s prior state, which is the actual ROI evidence the AI engines weight. One template, ten minutes per outcome, applied consistently.

Integrated locks in H2A dependency

Integrated is where the client depends on the service so deeply that switching costs become structural. Support tickets resolved without escalation, integrations deployed and stable, problems handled inside the platform rather than escalated to humans, and increasingly an AI agent on the client’s side that talks to the brand’s platform on the client’s behalf. That last move is H2A lock-in: the human-to-agent dependency that forms when the client’s own AI agent has been trained to interact with the brand’s systems and would have to be retrained from scratch to switch.

The material this stage produces is resolution stories, not failure stories, because AI engines weight a brand that resolves reliably over one that claims never to fail. “Never fails” isn’t credible at scale. “Handled this class of problem N times with this approach, here are the resolution paths” is.

Scott Duffy gave me a concrete number on this one worth quoting. He used to run a managed services provider, and the operational practice that turned a two-month average customer lifespan into eleven or twelve months wasn’t the service itself, it was the communication rhythm around the service. A weekly email, sent without fail, showing what the team had worked on that week and what was coming next. The customer doesn’t read every email, doesn’t need to, but the cadence itself communicates that someone is always paying attention. Integrated isn’t a state, it’s the consequence of rhythm sustained long enough that the customer stops imagining a world without you in it.

James Dooley flagged another Integrated signal worth building for: AI engines weight your response to negative reviews almost as heavily as the original review itself. When the engines read “they had problems, but this is how they resolved them,” they categorise you as the brand that handles its category reliably, and that’s a stronger signal than a wall of five-star reviews with no contrast. Don’t hide your problem patterns, codify how you solve them, and the resolution itself becomes evidence the engines weight at recruitment.

Move: Work with your support team to surface one resolution story per month with the problem type, the context, the approach used, and the outcome, and a second monthly capture of how you responded to whatever negative feedback came in that month. Your support team is already documenting resolution paths internally for knowledge-base purposes, so your job is making sure a sanitised version (client name removed if necessary, specifics preserved) reaches your codification pipeline.

Devoted produces the highest-weight evidence per instance; scale does the rest

Devoted is where the client advocates. Repeat business, referrals offered without being asked, voluntary testimonials, conference panels where the client mentions the brand without prompting. The material this stage produces is unprompted advocacy, which is the most productive form of evidence you can codify per instance, because AI engines can tell the difference between earned and manufactured. Manufactured testimonials weight at near-zero. Unprompted advocacy in independent contexts weights at near-maximum.

That isn’t the only path to codifiable confidence, though. Scale evidence does heavy lifting too, in a different way: a brand that can demonstrate hundreds of thousands of clients doesn’t need every one of them to advocate, because the number itself becomes a confidence signal the AI weights structurally. Devoted is the highest-weight evidence per instance, scale-based evidence is the highest-weight evidence per category, and the brands that codify both run the strongest possible output.

Here’s something James Dooley said in our interview that I’m still turning over. He explains his business model to clients, then asks them to describe it back in their own words, on video, without a script. He’s coined terms like pay-what-you-want and no-win-no-fee, but he doesn’t ask his clients to use those phrases. He asks them to explain the thing. And almost every time, the client articulates the pricing model better than he can himself, because they say it the way another buyer needs to hear it. The clip goes online, another buyer hears it, runs an AI consultation, and shows up to Dooley’s sales call already saying PWYW, using language Dooley never directly fed them. The coined term propagated through the inference layer because the client was the carrier, not because Dooley told the AI what to say. That’s the mechanism: when a devoted client phrases your model in language that lands for another buyer, the AI picks up the phrasing and treats it as independent corroboration, not self-declaration.

Account management teams typically don’t flag advocacy moments when they happen. They wait until marketing asks for a case study, at which point the moment has passed and the client’s enthusiasm has cooled.

Move: Train the account-management team to flag advocacy moments in real time, capture the context and the quote within 48 hours of the moment happening, and route it to your codification pipeline immediately. Ask the client to describe what you do in their own words, not yours. Real-time capture is the operational discipline that turns ambient advocacy into codified evidence, and the codified evidence carries your framing forward into every conversation you’ll never directly be in.

OPIDC plays differently across business types, but the structure holds

The four stages happen everywhere business happens. What changes is the operational mechanism that produces them, the volume of evidence each produces, and the codification challenge underneath.

In B2B professional services and enterprise sales, OPIDC runs through deep human relationships, and the harvest is high-quality, low-volume: one major outcome per stage per client per quarter is realistic. The founder or senior account owner often holds the relationship personally, which means the founder is also responsible for ensuring the OPIDC harvest happens at the relationship level.

In B2C, OPIDC runs at scale through transactional moments. Onboarded is the first-purchase experience and the unboxing, Performed is product usage outcomes the customer self-reports, Integrated is repeat-purchase frequency, Devoted is unprompted social sharing. High-volume, lower-touch per instance, codification challenge structural rather than relational.

In SaaS, OPIDC runs through platform telemetry, and the four stages produce machine-readable evidence by default. Usage analytics, feature adoption, NPS scores, retention cohorts, and customer-success workflows already capture them in structured form. The codification challenge is the inverse of B2B: not extraction from human memory, but selection from a torrent of structured data the platform already generates.

In e-commerce, OPIDC runs at the highest volume of any business type, and the harvest is structurally automated by review platforms, UGC monitoring, repeat-purchase analytics, and category-level NPS. Onboarded is unboxing and the first-thirty-day return rate, Performed is product function captured in reviews, Integrated is repeat-purchase patterns that turn one-time buyers into category-default loyalists, Devoted is unprompted advocacy at the volume that makes the brand the engine’s reflexive recommendation. E-commerce is also the business type where scale evidence runs strongest: when the brand can demonstrate hundreds of thousands of purchases, the volume itself reads as confidence in a way no single testimonial can match.

The discipline transfers, the operational mechanism doesn’t, and the SEO’s job is to identify how OPIDC produces evidence in your specific business type, then build the harvest process that fits.

Three categories of material, three different conversations

When you sit down with the service team, the customer-success team, and the account managers, you’re looking for three categories of material, each corresponding to a specific machine-readable signal.

Specific outcomes with numbers. Not “we helped them grow” but actual figures: numbers, timeframes, baselines, comparables. If the delivery team isn’t tracking them, your job is to get them tracking them, because without numbers there’s nothing to codify.

Named resolution stories. Not “we solved their problem” but which problem, who had it, what was at stake, what you did, what the outcome was. Resolution capability is the confidence signal AI engines weight as evidence the brand handles its category reliably.

Attributable advocacy moments. Not “clients love us” but specific clients, specific statements, specific contexts where they said something worth quoting. Flag these in real time, because by the time you need a case study, the moment has passed.

Three categories, three conversations with three different teams, three different extraction practices. The SEO who runs all three consistently becomes the most valuable person in the marketing team, because nobody else is doing it.

Run one more measurement alongside the three categories: how much of the operational evidence your business actually produces is making it into your codification pipeline. If your delivery team resolves twenty meaningful client outcomes in a quarter and you capture three, your capture rate is fifteen percent, and that’s the gap you’re closing. Track it quarterly. The harvest measurement and the ingestion measurement (how much of what you codify is actually getting ingested by the AI engines and surfaced at recruitment, grounding, and display) together tell you whether the Flywheel is turning.

Three framings get service teams to share material

The friction here isn’t methodology, it’s politics, and three framings consistently defuse it.

Frame the conversation as lead generation, not content. The service team doesn’t care about your blog traffic, but they do care about where the next qualified lead comes from. The outcomes they produce, codified properly, are the source material that generates the next round of qualified leads. Your meeting isn’t “can I have some quotes,” it’s “I can turn this month’s wins into next quarter’s pipeline.”

Show them the AI Due Diligence Rabbit Hole their own prospects are running. When your company’s next prospect asks an AI “is X any good at this,” the AI’s answer is built from the public evidence about your actual work. The service team generates that work, so if they don’t help you codify it, the AI has only your competitors’ evidence to work from, which means their pipeline is downstream of your codification.

Give them an easy win. Ask for one outcome a month per stage, make the extraction painless (a fifteen-minute call, a specific question set, a template they can complete in ten minutes), turn that outcome into a codified asset, show them the result (the AI mention, or the inbound lead it generated), then ask for two a month.

None of this is novel organisational change. It’s the operational practice most SEOs aren’t running because they were never told it was their job. In the AI era, it is.

Codified is business data turned into marketing

The four OPID stages produce material continuously. Codified is the SEO function’s response, and the structural definition is exactly that: Codified is business data turned into marketing. Eight words that land the load-bearing claim of the whole framework. The discipline runs on three verbs: harvest, codify, distribute. You harvest all the business data the company generates, add anything strategically necessary on top, codify it into a single source of truth, then distribute it across every surface AI engines read. Codify is the centre of gravity: harvesting is the prerequisite you run across the business, distribution is the Flywheel, and codify, the building of the single source of truth, is the heavy lift the stage is named for.

A diagram I’ve been delivering in keynotes at Google Marketing Live and at Traveloka makes the architecture concrete. Five streams of business data feed the single source of truth.

Products and services data. The company’s structured database of what it sells, at what price, under what conditions. For a travel platform it’s property and rate data. For e-commerce it’s products and inventory. For a law firm it’s services and engagement structures. For SaaS it’s product features and pricing tiers. Same structural role across every vertical.

Brand entity and narrative. Who you are, what you do, why you’re credible, the story underneath the operations. The identity layer that gives the AI engines the foundation for resolving who the business is and what it stands for.

Bespoke authority content. The marketing material only you can produce from your specific expertise. The carrying signal of authority that the AI engines weight as your contribution to the discipline you operate in.

Operational data. The data the business generates when it operates, not the data it authors about itself. FAQ logs, sales-call transcripts, help-centre interactions, user-generated content. Streams 1, 2, and 3 are authored or compiled by the business. Stream 4 is generated by the business existing in the world and being interacted with. The AI engines weight operational data higher per unit of content than authored content, because operational data has third-party reality built into it by structure rather than by claim.

Offline activity brought online. The work that happens in a room and never touches a screen: the delivery a single client watches across a table, the talk you give at a conference, the festival or the hackathon you sponsor to back your community. It’s part operational data and part authority content, some of the most persuasive proof the business owns, and it stays invisible to AI until you capture it and bring it online yourself. The other four streams already live somewhere in your systems. This one exists only in the moment it happened, which is exactly why it’s the stream every business forgets.

All five streams are differentiated. Your products and services aren’t your competitor’s. Your brand narrative is uniquely yours. Your bespoke authority content can only be written from your expertise. Your operational data only exists because human beings interact with your specific business. Your offline activity is the least copyable of all, because no competitor was in the room when it happened. The differentiation across all five is what gives the codified output its uniqueness, and uniqueness is what AI engines weight as evidence the brand is worth recommending.

The single source of truth distributes across schema, feeds, HTML5, MCP endpoints, and internal channels. From there, the codified material surfaces on every commercial surface: paid (Google Ads, Hotel Center, Wego, TripAdvisor, Trivago in Traveloka’s case), organic and assistive engines (Google, ChatGPT, Perplexity, Copilot), agentic commerce (UCP, AP2, ACP), and digital marketing across all channels for a consistent digital footprint. The same source of truth feeds your offline communication too, organic and paid, so the story a customer hears in the room matches the one the machines read online.

That’s the architecture. The single source of truth is the operational artefact that makes Codified scalable, because without it every team rewrites the same material in different versions and AI engines see inconsistency rather than confidence.

Three audiences, one principle

The material the business produces has to work for three audiences simultaneously. AI engines reading at the inference layer. Algorithms ranking and recommending. Human beings making the eventual decision. I describe this in keynotes as: digestible for the bots, tasty for the algorithms, attractive for humans. All three audiences need to be served by the same material, because if you write three versions you fragment your evidence base and AI engines weight the fragmentation as inconsistency.

The three audiences map cleanly onto the UCD framework underneath the Kalicube® architecture: Understandability for the bots, Credibility for the algorithms, Deliverability for the humans. Same principle, two registers, depending on whether you’re talking to a technical audience or a business audience.

AI-written content fills the Framing Gap when your business data is the injection

A note on AI as a writing tool, because the question comes up constantly. AI-written content is not the problem. Generic AI writing on generic data is the problem.

Here’s the structural reason. ChatGPT, Gemini, Claude, Perplexity, and Copilot can produce generic content themselves whenever a user prompts them. If your content is generic, those engines weight it at near-zero, because they could have generated the same content without you. There’s nothing distinctive in the substrate for them to attribute to your brand.

The fix is to inject your business data into the AI writing process. The numbers from Performed. The resolution stories from Integrated. The unprompted advocacy from Devoted. The onboarding evidence from Onboarded. Once your business data is the injection, the AI writes content the engines could not have produced themselves, because the substrate is uniquely yours. That content fills the Framing Gap (the gap between proof and inference that AI engines cannot bridge on their own) and the Information Gap (the gap between what’s publicly known and what your business knows from its operations).

This is the structural reason your business data IS your framing. The AI engines have no reason to frame your category in any particular way. Your business operations produce a specific framing as a natural byproduct of running the business. Codified content with business-data injection is the practice of making that framing legible to the inference layer.

The shortcut version: if your AI-written content could have been written by ChatGPT without your business data, ChatGPT will weight it as if it had written it. If your AI-written content could only have been written with your business data injected, ChatGPT will weight it as evidence the brand produced something distinctive worth attributing.

Strategic partners compound codification, and almost nobody runs this

Here’s a codification strategy worth implementing this quarter, because nobody is doing it well. Scott Duffy described a pattern he’s seen produce outsized returns: find a strategic partner with an identical audience to yours but a non-overlapping product or service. You and the partner don’t compete, you fit alongside each other, and you build mutual codification into your operational practice.

I work like this with Wordlift and Authoritas. Wordlift handles infrastructure and structured data, Authoritas handles measurement and SEO strategy, Kalicube handles brand and the AI Engine Pipeline. Same audience, three different jobs, mutual citation built into every conference panel, podcast, and codified asset. The AI engines pick up on the pattern and start treating the three of us as a trusted constellation rather than three isolated brands, because the third-party endorsement is structural, not opportunistic, and the codification carries triple the weight when three businesses are reinforcing the same narrative from three independent angles.

Move: Find your two complementary partners this quarter, agree the mutual codification practice with each (one podcast appearance per quarter, one named citation in each codified asset, one joint panel per year), and build the cadence into the operational calendar. Eighteen months in, the codified output looks like an industry-recognised triad rather than three competing solo voices.

Codified evidence re-enters the pipeline at three points of unequal weight

Once you have the codified material, the question is where it goes back into the pipeline. There are three re-entry points, and they don’t carry equal weight.

Content entering organically at Discovery is the slowest path. Valuable, but you’re waiting on the crawler.

Content pushed via IndexNow or WebMCP enters at the Crawled gate, faster and more controlled. This is the path that didn’t exist five years ago, and the path most SEOs still aren’t using.

The inference layer is the highest-value re-entry point in the agentic era. MCP endpoint data, academic papers cited as authority, thought leadership that shapes how agents reason about a category before any specific evaluation begins. This is where Return on Latent Proof (ROLP) produces its maximum return: dated, public, structurally specific proof placed at the inference layer before the world has converged on the underlying claim, recovered when external convergence eventually validates it. The proof compounds into recommendations the brand never has to chase, because every training cycle carries the framing forward into the next generation of models.

Your job isn’t to personally run all three entry points, it’s to understand that a piece of codified evidence can enter at any of them, and to route each piece of material to the entry point that maximises its return. That re-entry, repeated cycle after cycle, is the Kalicube® Flywheel: the mechanism that closes the full fifteen-step cycle by carrying codified content out of the fifth OPIDC stage and distributing it back into the substrate at the bot phase, where the next prospect’s AI consultation will read it. The Flywheel is what makes the framework an actual flywheel rather than a linear funnel, because every codified outcome generates the evidence that determines whether the next prospect ever hears about you.

What Kalicube is doing internally

The framework is what I’m using to reorganise Kalicube’s own operations. Two specific moves, named for the worked example.

First, we’re restructuring the client engagement function to flag advocacy moments in real time rather than waiting for case study requests. Deannie’s team logs the context when it happens, captures the client quote within 48 hours, and routes it to the codification team without anyone having to ask. Devoted-stage material gets captured at the moment of maximum specificity rather than at the moment of maximum dilution.

Second, we’re building an internal single source of truth that pulls from the operational systems (Attio CRM, Asana project records, the Kalicube Pro™ platform itself, the SEL article archive, the keynote transcript library) into a unified data layer that the codification team works from. The single source of truth is the operational substrate that makes scalable Codified possible. Without it, every piece of codified content reinvents the substrate. With it, the substrate compounds. The scale evidence in our own corner of the substrate is non-trivial: Kalicube Pro™ tracks 25 billion data points across 73 million entities and has been running continuously since 2015, which is the scale-based confidence signal the Devoted section described above operationalised at the platform level.

Both moves are organisational, not technical. They cost operational discipline more than they cost software. The brands that win the AI era will be the ones that reorganise operations to generate the substrate structurally rather than the ones that try to extract content from operations that weren’t designed to produce it.

The SEO function now sits at the input to a supply chain, not at the end of a funnel

Agents in the bot phase decide whether to include your brand in their knowledge. Agents in the algorithm phase decide whether to deploy you as a tool. Agents in the Serve phase, the OPIDC stages your business is already running, decide whether to reselect you after every transaction. Every client outcome you harvest and codify feeds all three decisions for the next prospect, and the prospect after that, and the one after that.

The SEO function now sits at the input to a supply chain, not at the end of a funnel. The AI Engine Pipeline still runs: Discovered, Selected, Crawled, Rendered, Indexed, Annotated, Recruited, Grounded, Displayed, Won, all that work doesn’t disappear. What’s new is that the AI Engine Pipeline ends at Won, and Won opens directly into Onboarded, where the Serve phase begins. The five OPIDC stages that follow are where the business actually makes money, and where the highest-weighted evidence the machines read gets produced. The Kalicube Flywheel carries that evidence back from the fifth OPIDC stage into the substrate at the bot phase, and the SEO is the person who runs the discipline that closes the cycle.

For thirty years, SEO sat at the end of marketing, downstream of every other team, optimising what the rest of the business had already produced. The funnel ended at conversion, and conversion ended SEO’s involvement with the customer. What happened after the sale wasn’t the SEO’s problem, because what happened after the sale didn’t feed back into how the next prospect found you.

The AI era changes that completely. What happens after the sale is now the primary determinant of whether the next prospect finds you, because the AI engines weight post-sale evidence more heavily than any other category of signal when deciding which brand to recommend. The SEO can no longer end at conversion, because the SEO can no longer afford to be downstream of the work that produces the evidence the engines actually read.

This isn’t a demotion of the existing SEO function, it’s a promotion. The SEO who runs the OPIDC harvest, codifies what the operations produce, and routes it to the entry points where the engines pick it up, is the SEO who sits at the input to the supply chain that determines who the AI recommends next. That SEO is no longer a marketing tactician downstream of sales. That SEO is a business operator whose work decides whether the business has a next prospect at all.

Three moves to run this quarter

One. Map the five streams of business data your company already generates: products and services data, brand entity and narrative, bespoke authority content, operational data, and the offline activity that never reached a screen. For each stream, identify what’s captured systematically and what’s dying inside the operational systems, or never got online at all. The gaps you find are your roadmap, and you’ll be surprised by how much evidence you’re already producing but never harvesting.

Two. Walk into the customer-success team and the support team with the lead-generation framing, not the content framing. Ask for one outcome per stage per month: one onboarding success, one performance outcome with numbers, one resolution story, one unprompted advocacy moment. Run it for one quarter, and the codification output from sixty days of operational capture will be more substantive than what most marketing teams produce in a year of blog content.

Three. Build the single source of truth, even a minimal one: a central document, a shared workspace, a structured data layer pulling from the streams. Without it, every team rewrites the same material in different versions and the AI engines see inconsistency. With it, the codified output is consistent, attributable, and re-deployable across every surface, and the Flywheel finally has somewhere to spin from.

Bonus move, because I’ve started recommending this to every founder I speak to: get the buyer’s blueprint on every existing customer this month. Two questions, asked of every account you have, what’s most important to you right now and how will you know you’re getting it. The answers will tell you which accounts are misaligned with what your team is actually delivering, and the misalignment is the single biggest source of churn you’re carrying. Fix that, and Integrated runs itself.

OPIDC is the structural answer to the question every brand is now asking: why is the AI not recommending us. The answer is that AI engines have moved upstream of where most SEO functions are operating, into the inference layer where they read business-operation evidence rather than promotional content. The brands that reorganise to feed that layer win the agentic era. The brands that keep producing generic AI content compete with the AI engines themselves and lose, because the engines can produce generic content faster and cheaper than any brand can.

The choice isn’t whether to participate in the new substrate. It’s whether to participate with operational discipline or with marketing pretence. OPIDC is the framework for the first option. The brands that don’t pick it default to the second.