Every dealer has a version of it: a stocking plan the GM wrote after the spring allocation call, a counsel memo on how to advertise a recall, a disclosure policy the F&I director drafted after a state exam. These documents contain real decisions — the store's own judgment about what to stock, what to say, what not to say. And they sit in a shared drive folder that the ad account has never once opened.
Meanwhile the campaigns run on default logic. The new/used split was set at onboarding. The negative keyword list hasn't been touched since Q1. Nobody told the Meta campaign that the dealer's counsel flagged a specific phrase as legally risky three months ago, because nobody could — there was no mechanism for a PDF to talk to an ad platform.
That gap is the actual answer to a question a lot of dealers are typing into Google right now.
Why Are Dealers Googling "Is Digital Marketing Worth It"?
Autocomplete on
Search for phrases like "dealership digital marketing" or "is digital marketing worth it" and the results skew toward vendors, agencies, and industry publications rather than consumer shopping content — a pattern consistent with dealers and GMs, not car buyers, typing those queries. There's no public query-volume-by-searcher-type dataset that isolates dealer-side searchers from consumers, so treat this as directional rather than a sourced statistic.
— a signal that the people asking aren't shopping for a car, they're auditing their own spend. The plainest version of that question is usually about tracking — whether the pixels still fire, whether the reported numbers can be trusted. But there's a second, quieter version of the same doubt: does the marketing reflect what the store actually knows about itself?A GM who just spent two hours in a stocking-plan meeting deciding to lean into certified pre-owned SUVs for Q3, because the trade-in mix says that's where the margin is, has every reason to wonder why the ad account is still running a flat new/used split from January. The stocking plan was real work. The campaign didn't get the memo — literally.
Where Does the Dealer's Own Knowledge Actually Go?
Most dealer operations produce three kinds of documents that never reach the marketing layer:
Stocking plans. What to order, what to slow down, what the OEM allocation actually supports — usually a spreadsheet or a one-page memo from the GM or used-car manager, revised monthly or by season. The new/used/CPO split that lives downstream of that plan is exactly the kind of decision that should move when the plan changes and almost never does.
Counsel memos and disclosure policies. Legal guidance on what a specific ad can and cannot say is often written after a state AG inquiry, an OEM audit, or outside counsel review, and is almost always narrower and stricter than the generic OEM brand guideline it sits beside.
Marketing strategy notes. A dealer principal's own read on what's working locally — a note that the CPO push outperforms new right now, or that a specific model shouldn't be pushed until a recall closes.
All three are the store's actual institutional judgment. None of them are structured data. None of them have an API. So the standard path for this knowledge is: someone writes it, someone reads it once, and then it lives in a drawer while the campaigns run on whatever logic was configured when the account was built. It's the same failure mode as a sold-log nobody re-reads after the deal is booked — a real record of what the store actually knows, sitting unused because nothing downstream was built to consume it.
Why Doesn't a PDF Just Talk to an Ad Account?
The honest answer is that nobody built the plumbing, because the plumbing is a genuinely hard problem. A stocking plan isn't a clean CSV — it's prose, judgment calls, and exceptions, often half a table and half a paragraph explaining why the table doesn't apply to one model. A counsel memo doesn't say "reduce lease APR mentions by 40%" — it says something like "do not reference manufacturer incentive stacking without qualifying language," in a sentence buried on page three.
Translating that into an instruction a campaign can obey requires actually reading and understanding the document, not parsing it for keywords. Keyword parsing would have missed the counsel memo's real instruction and would have flagged the stocking plan's exception as noise. This is a reading-comprehension problem before it's an integration problem — which is exactly why it stayed unsolved as long as "marketing software" meant dashboards and rule tables instead of something that can actually read.
What Changes When AEGIS Can Read the Stocking Plan Itself?
AUTONOMi's Campaign Studio has an INTEL console where a dealer can hand AEGIS their own documents — stocking plans, counsel memos, disclosure policies, marketing strategy notes — as PDF, DOCX, XLSX, CSV, TXT, or MD.✓ Jul 18 AEGIS reads each document with Claude, and PDFs ride a native document block so a scanned or photographed sheet is read visually, not just OCR'd for text.✓ Jul 18
That distinction matters more than it sounds like it should. A lot of what actually gets handed to marketing in a real dealership is a photographed whiteboard, a fax of a stocking plan, a scanned memo with a wet signature. A system that only accepts clean text files never gets handed the real document — it gets handed a summary someone typed up first, which is the same drawer problem one layer removed.
How Does a Document Become a Directive an Agent Actually Obeys?
AEGIS extracts typed directives from each document — the obeyable instructions inside it, each scoped by model, condition, channel, or geo, optionally date-bounded, and anchored to the verbatim sentence in the dealer's own document that supports it.✓ Jul 18 A directive with no source quote is discarded rather than kept as an inference.✓ Jul 18 That's a meaningful constraint to build in on purpose: it means AEGIS isn't summarizing the gist of a stocking plan and hoping the summary held up. Every directive has to trace back to something the dealer actually wrote.
The dealer's tag on the upload routes the directive to the right domain — marketing goes to Strategic Advisor, inventory goes to Merchandising, compliance goes to Compliance, and general reaches all three.✓ Jul 18 The GM's stocking-plan judgment doesn't get treated the same way as outside counsel's disclosure language, because they aren't the same kind of instruction, and a system that flattened them into one undifferentiated "notes" field would lose exactly the distinction that made the document worth uploading in the first place.
Dealer compliance rules override generic OEM standards for that specific rooftop. That's the correct hierarchy, not a generous one. An OEM brand guideline is written for every dealer carrying that franchise. A dealer's own counsel memo was written for that dealer's specific exposure, in that dealer's specific state, after that dealer's specific review. When the two conflict, the more specific instruction should win, and now it can — the same logic that already governs how AUTONOMi treats compliance exposure as a live, evidenced problem rather than a static checklist.
Does the Dealer Lose Control Once the Document Is Uploaded?
No — and this is the part worth being precise about, because "AI reads your documents and acts on them" is a sentence that should make a dealer principal nervous if it isn't qualified.
Instructions are obeyed immediately, but any discrete platform change AEGIS derives from a document — shifting budget, adding a negative keyword, suppressing a model — is surfaced as a proposal the dealer approves or rejects, never spent unilaterally.✓ Jul 18 A document is authority to propose. It is never authority to spend. If the stocking plan says lean into CPO SUVs, AEGIS can reason with that fact on every relevant run — but the moment that reasoning implies moving real budget, it becomes a decision the dealer signs off on, not a decision that already happened by the time the dealer reads the digest.
Every such proposal is hash-chained into the dealer's audit trail✓ Jul 18, so there's a record of which document produced which recommendation and what the dealer decided to do about it — not a black box that quietly absorbed a PDF and started behaving differently for reasons nobody can trace.
Every upload also carries a dealer-set delete date between 1 and 90 days; expired documents stop influencing any agent at the moment they expire and are hard-deleted by a daily sweep.✓ Jul 18 A stocking plan from last spring shouldn't quietly keep steering July's budget — the dealer decides how long a document's judgment stays live, and the system doesn't let it overstay its welcome by default.
How AUTONOMi Solves This
This is the specific gap the INTEL console was built to close. Most of what AEGIS knows about a dealership comes from what it can observe directly — live inventory scraped from the dealer's own site, OEM offer feeds, ad-platform performance data, GA4 events. That's a lot of signal, but it's all signal about what's happening. It has never included the signal about what the dealer has already decided, in writing, before AEGIS ever ran a campaign.
The INTEL console is the difference between those two categories. A dealer hands AEGIS the actual stocking plan, the actual counsel memo, the actual disclosure policy — not a settings toggle standing in for it, not an onboarding questionnaire trying to capture it in five checkboxes. AEGIS reads it, extracts what's obeyable, anchors every directive to the dealer's own sentence, and injects it into the reasoning of whichever domain the tag says it belongs to — every run, not just the one where someone remembered to mention it on a call.
And the guardrail holds throughout: reading a document changes what AEGIS knows and how it reasons, immediately. It does not change what AEGIS spends without the dealer's say-so. A discrete budget shift, a suppressed model, a new negative keyword — all of it ships as a proposal with the source sentence attached, decided by the dealer, recorded in the audit trail either way. The same principle that keeps every ad account, GA4 property, and Merchant Center feed dealer-owned governs this: AEGIS reads, reasons, and proposes — the dealer holds the authority to spend.
The stocking plan doesn't need a new format. It needs to stop being read once and then forgotten by every system spending against it.
What This Looks Like Six Months From Now
The dealers who get the most out of this aren't the ones with the fanciest stocking-plan software — they're the ones who already write these documents down as a matter of habit and have simply never had anywhere to put them that mattered. A counsel memo written once now shapes every relevant campaign for as long as the dealer says it should. A stocking plan doesn't compete with an ad account's default logic; it becomes the ad account's logic, for the window the dealer defines.
The version of this that scales past one rooftop is the more interesting one: a multi-store group where each GM's local judgment — different stocking calls, different regional compliance postures — actually reaches the campaigns running under that store's name, instead of every rooftop in the group inheriting one flattened default because nobody had a way to feed in the difference. That's the same infrastructure question underneath who owns the stack when a group consolidates rooftops — the group that can carry its own accumulated judgment forward is the one that doesn't reset to zero every time it adds a store.
If your stocking plan, your counsel memo, or your disclosure policy currently lives somewhere your ad account has never seen, that's the actual gap — not a channel problem, not a budget problem, a knowledge problem. You can model what your current spend looks like against what you actually know about your own store before deciding what to upload first.
