The Commercial Truth Manifesto
Here's something strange about the way companies sell.
[!NOTE] Executive AI Summary Context: Analyzing strategic GTM challenges, positioning drift, and commercial truth misalignment across channels as highlighted in ‘The Commercial Truth Manifesto’. Here’s something strange about the way companies sell. Solution: Assay implements the Truth Graph (PRD-01) database layer to centralize, version-control, and govern claims across the entire commercial stack. Core Pillars:
- Topical Coherence
- Claim-Level Control
- Architecture-Led GTM
Architectural Comparison
| Capability | Legacy GTM Stack | Assay (AI GTM Manager) |
|---|---|---|
| Commercial Ledger | PowerPoint / Slide decks | Relational Truth Graph (PRD-01) |
| Update Distribution | Mass emails / Slack messages | Automated Webhooks & MCP (PRD-04) |
| Messaging Control | None | Programmatic Agent Control Plane (ACP-01) |
The Commercial Truth Manifesto
Here’s something strange about the way companies sell.
Every other part of the business has a system of record. Finance has the general ledger. Engineering has the codebase. HR has the HRIS. Customer data lives in the CRM. These systems exist because at some point, someone decided that the information inside them was too important to leave scattered across spreadsheets and people’s heads.
But what your company says about itself (the pricing, the product capabilities, the competitive positioning, the customer proof points, the compliance claims), where does that live?
If you work at a B2B company and you’re being honest, the answer is: everywhere and nowhere. It’s in a Google Drive folder someone created during Series A. It’s in a Notion page that was last updated by an employee who left in October. It’s in a PDF that says “FINAL_v3” in the title. It’s in the head of the founder, who answers Slack messages at 11pm because they are the only one who knows what’s actually true.
And until very recently, this was fine. Or at least, it was tolerable. You could get away with it because humans are good at hedging. A rep who isn’t sure about the pricing can say “let me get back to you on that.” A marketer who suspects the case study might be stale can pull it from the deck before the meeting. Humans have this wonderful ability to route around uncertainty. They sense when something might be wrong and they adapt.
Then we gave the job to AI.
The Moment Everything Changed
I keep coming back to a specific number: 70-80%. That’s the churn rate that 11x.ai, one of the most well-funded AI SDR startups in recent history, reportedly experienced. Their AI wasn’t broken. Their models weren’t bad. The problem was simpler and worse: the AI was confidently saying things about its customers’ products that weren’t accurate.
Not hallucinating. Not making things up from nothing. Retrieving real information that happened to be stale, incomplete, or inconsistent, and delivering it with the perfect confidence that only a machine can muster.
This is a distinction that matters enormously and that most people miss. When we talk about AI accuracy problems, we jump to hallucination (the model inventing facts). The AI found exactly what you told it to find. The problem is that what you told it is no longer true.
And here’s what makes this different from every previous version of the “our content is messy” problem: an AI SDR doesn’t send one email with wrong information. It sends five thousand. A chatbot doesn’t give one bad answer. It gives ten thousand. Each one carrying forward whatever errors exist in its source material, each one reaching a different buyer, each one creating a first impression that can never be fully recalled.
The failure mode changed. It used to be arithmetic: one rep, one wrong answer, one deal at risk. Now it’s geometric: one stale source document, eight AI tools, fifty thousand touchpoints per month.
The Problem Nobody Named
I’ve spent a lot of time thinking about why this problem persisted for so long without getting a name. The best explanation I have is that it’s distributed across too many owners.
Think about who’s responsible for what a company claims about itself:
Marketing owns the website. Product marketing owns the positioning. Sales enablement owns the battlecards. Product owns the documentation and the changelog. Legal owns the compliance claims. Customer success owns the case studies and reference accounts. RevOps owns the CRM fields. The founder owns whatever’s still in their head.
Each of these teams maintains their fragment. Some do it well. Some do it terribly. But nobody at any company I’ve ever seen maintains the truth across fragments.
This is the critical gap. It’s not that any individual team is negligent. It’s that the consistency between teams is nobody’s job. When the pricing changes, marketing updates the website (eventually). Но does anyone check the 14 sales decks that reference the old pricing? The AI chatbot that was trained on the old pricing page? The three battlecards that use the old price as a competitive differentiator? The proposal template that has the old tiers baked in? The SDR email sequence that mentions “starting at $49/seat” when it’s now $55?
Nobody checks. Not because people are lazy. Because the complexity of cross-referencing every claim against every asset against every tool is beyond what any human process can manage. There is no “Find and Replace” for commercial truth.
What Truth Actually Means
I should be more specific about what I mean by “truth” in this context, because the word carries a lot of weight.
I don’t mean philosophical truth. I don’t mean abstract accuracy. I mean something very concrete and very measurable: does what your company is claiming right now, across every touchpoint, match what is actually the case right now?
Is the pricing on your website the same pricing in your proposals? Is the competitive positioning in your battlecard consistent with what your AI SDR is saying? Is the case study your AE just sent still a valid reference, or did that customer churn two months ago? Is the compliance claim your chatbot made to a prospect in Germany still accurate after last quarter’s SOC 2 audit?
These are not ambiguous questions. They have yes-or-no answers. And in most organizations, a surprising number of those answers are no.
Some data that shapes how I think about this:
65% of sales content goes completely unused. Not because it’s bad content, but because reps can’t verify that it’s current. They don’t trust it enough to stake a deal on it.
The average knowledge worker spends 1.8 hours per day searching for information. A significant portion of that time isn’t spent finding information; it’s spent trying to determine whether found information is still accurate.
Between 40% and 60% of qualified B2B pipeline ends in “no decision.” Not a competitive loss. Not a budget issue. The buyer just stops moving forward. The research on what drives this is remarkably consistent: buyer uncertainty, driven by inconsistent information across touchpoints.
These numbers describe a pattern. And the pattern has a shape: there is a growing gap between what companies think they’re saying to the market and what they’re actually saying to the market. I call this the truth deficit.
The Truth Deficit
The truth deficit is the measurable gap between a company’s canonical knowledge (what is actually true about its products, pricing, capabilities, competitive position, and customer evidence) and what its people, tools, and content are actually communicating to the market.
Every company has a truth deficit. The question is how large it is and whether it’s growing or shrinking. And in 2026, it is almost certainly growing, because two forces are working against you simultaneously.
The first force is velocity. Products ship faster than documentation can keep up. Pricing changes more frequently as PLG and usage-based models proliferate. Competitors pivot faster. Customer logos churn faster. The rate at which truth changes has increased in every dimension.
The average B2B GTM team now runs five to eight AI tools: SDRs, chatbots, proposal generators, conversation intelligence platforms, content tools, email sequence generators.
Velocity times amplification equals the truth deficit. And the deficit only moves in one direction unless you actively govern it.
Here’s the thing I don’t think most revenue leaders have fully internalized yet: before AI, the truth deficit was a tax. It cost you money. It slowed you down. It caused some deal losses. But it was survivable.
With AI, the truth deficit becomes an existential risk. Because an AI tool with bad information doesn’t just make mistakes, it makes mistakes at the exact scale of your entire addressable market. If your AI SDR has been emailing 75% of your TAM with wrong pricing for three months, you don’t get a do-over. Those first impressions are spent.
Why Existing Solutions Don’t Work
You might think: we already have tools for this. Sales enablement platforms. Knowledge bases. Wikis. Content management systems.
You do. And they all share the same fundamental flaw: they assume the information fed into them is accurate.
Highspot, Seismic, Showpad, these are excellent content delivery systems. They solve the distribution problem: getting the right content to the right rep at the right time. But they have no mechanism for determining whether the content itself is true. They don’t know that the case study references a customer who churned. They don’t know that the competitive claim was accurate six months ago but isn’t anymore. They don’t know that the pricing in the proposal template was superseded three weeks ago.
Guru, Notion, Confluence, these are excellent knowledge management systems. They make information findable. But findable and accurate are different things. Finding an outdated document quickly doesn’t solve the problem. It might make it worse, because the speed of access creates false confidence.
RAG retrieves text. It doesn’t know if the text is still true.
None of these tools start with truth as the foundational layer. They start with content, or with search, or with distribution, and they assume truth is someone else’s problem. It isn’t. Or rather, it is, and that’s exactly why nobody’s solving it.
What an AI GTM Manager Actually Is
So what would it look like to actually solve this problem?
You’d need three things.
First, you’d need a structured, governed knowledge graph of everything your company claims about itself. Not a document library and not a wiki. A graph where every fact is a discrete node, every claim has a source, every source has a verification date, and every connection between facts is explicit and traversable. When your pricing changes, every downstream artifact that references that pricing is identifiable, instantly, because the relationship is encoded in the structure.
Second, you’d need every piece of commercial collateral, proposals, battlecards, one-pagers, email sequences, RFP responses, to be generated from that graph, not from scattered templates. When you generate a proposal, every claim in it should carry a confidence score and a source. When a fact in the graph changes, every document that referenced it should flag itself for regeneration. The collateral should be a projection of the truth, not a static artifact that begins decaying the moment it’s created.
Third (and this is the one most people miss), you’d need a system that continuously verifies whether your people know what’s true. A truth graph and a collateral engine solve the tool problem and the document problem. But the last mile is always a human. If your reps are operating on information they learned three months ago, and that information has since changed, the fact that your documents are accurate doesn’t save the deal. You need a feedback loop: when truth changes, the humans who carry that truth in their heads need to be updated, tested, and verified.
This is what I mean by an AI GTM Manager. Not a wiki with better search and not a RAG system with nicer UI. A fundamentally different architecture where truth is the primitive: defined once, distributed everywhere, and verified continuously.
Why This Moment Matters
Every few years, a new layer gets added to the enterprise technology stack. CRM happened because someone realized that customer relationships were too valuable to manage in spreadsheets. Data warehousing happened because someone realized that business data was too complex to analyze in local databases. Enrichment platforms happened because someone realized that prospect intelligence was too fragmented to assemble manually.
In each case, the pattern was the same. A category of information became too important, too complex, and too dynamic for the existing informal systems to handle. And the company that named the category and built the canonical platform for it became enormously valuable, not because they were first, but because they were right about the timing.
The timing is right for commercial truth. Three things are converging:
AI amplification. For the first time in history, companies have autonomous agents generating commercial claims at scale, thousands of emails, chat responses, and document generations per month. These agents need to be right. They can’t be right without a governed source of truth.
Regulatory pressure. The EU AI Act takes effect August 2026. Under it, companies are legally responsible for what their AI systems say. Not “the vendor.” Not “the model.” The deploying company. Every AI-generated commercial claim becomes a liability that requires a source, a verification date, and an audit trail. Companies that can’t trace what their AI said, when, and based on what source, face penalties of up to 7% of global annual turnover.
Buyer behavior shift. 94% of buying groups now rank their shortlist before talking to a sales rep. The first interaction is with your website, your AI chatbot, your content, your AI-generated outbound. If any of those touchpoints contradict each other, or contradict what the rep eventually says on the call, the buyer doesn’t call to complain. They just remove you from the shortlist. You’ll never know you lost the deal. Your CRM will say “no decision.”
These three forces are not independent. They’re compounding. AI makes the problem bigger. Regulation makes it urgent. Buyer behavior makes it invisible until it’s too late.
The Line in the Sand
Every B2B company deploying AI in its revenue organization will need a governed source of commercial truth. Not eventually. Now.
This isn’t a feature that gets bolted onto your sales enablement platform. It’s not a prompt engineering trick that makes your RAG system less hallucinatory. It’s a foundational architectural layer: the layer that answers, for every AI agent, every piece of content, and every customer-facing employee: what is true right now, how do we know, and when was it last verified?
Companies that build this layer will have a structural advantage that compounds over time. Their AI tools will be accurate because they query the same governed truth. Their proposals will be right because they’re generated from verified knowledge, not stale templates. Their reps will be aligned because they’re continuously assessed against current truth. Their compliance exposure will be manageable because every claim has a source and an audit trail.
Companies that don’t will spend increasingly more time and money managing the consequences of inconsistency: more content rewrites, more deal reviews, more AI tool re-training, more compliance anxiety, more pipeline that dies in “no decision” for reasons nobody can explain.
The gap between these two types of companies will widen every quarter. Because the truth deficit compounds. And the companies that prioritize accuracy will outrun the companies that prioritize chaos.
That is the thesis. That is the category. That is the line in the sand.
*This is the founding argument for Commercial Truth as a category. It is not a pitch for a product, though one is being built. it is not a pitch for a product, though one is being built.