The era of action is here, and it arrived very quickly.
The yet-to-be-fully-realized promise of AI is the same for every executive: companies will grow what they deliver without growing what they cost. Tomasz Tunguz captured the shift in early 2026 — AI-native companies like Anthropic, Cursor, and Midjourney are running at $2M–$5M in revenue per employee, against a traditional SaaS benchmark closer to $200K–$300K. A 10–20x gap. As he put it, once competitors demonstrate this efficiency, it's untenable not to match it. The pressure is now on every executive's desk.
Buried inside that efficiency story is a question most companies are skipping past: efficient at what? An agent drafting emails twice as fast still costs you money if the emails were the wrong work to begin with. The AI-native companies above aren't just faster — they were built and commercialized around the outcomes their customers actually came for, and the efficiency followed from that. For an established company, retooling AI without that clarity just produces a faster version of work that may or may not have been worth doing. The question to ask before "how do we go faster?" is: do we actually know this is the best action to take to drive customer outcomes?
Action is only as good as the context underneath it, and everyone in the AI era has felt this firsthand — a moment of absolute magic followed by the realization that a deep research project made up every data point in the article, the demo that worked once and never again, the never-ending loop of "that actually sharpens things significantly" when iterating on the same idea. AI will always act instantly and with unmatched conviction, which makes it difficult to decipher whether the action is actually correct.
In the world of Customer Success for B2B SaaS, the right action means knowing three things — and most companies have never had a system that could answer any of them.
How your business operates. The most established in theory: every onboarding deck, playbook, "this is how we do things" training, and required field in the CRM or CSP. But the most fragile in practice — guidelines shared once at hire and rarely revisited, rigid workflows in systems of record that nobody actually follows. Systems of record give surface-level visibility into activity, but the more important question has always been: does the process actually reflect the next best action to deliver outcomes? That answer is complex and nuanced, and no system has been able to produce it.
How you deliver value to your customers. The best-documented externally: every company has a website, a sales motion, a value proposition, and case studies. But the most one-directional and anecdotal in nature — we describe what we do, but we've never had real-time feedback loops to validate whether the value we promise is actually landing. The validation question has always been: are we delivering what we said we'd deliver? No system has been able to answer that continuously, at scale, across every customer.
How each customer is realizing that value. The most elusive of the three. The CS industry has built health scores, NPS, and adoption metrics, but in reality these were always proxies for a truth nobody could capture directly. The data question has always been: are we measuring what's easy, or what actually matters? Value is more than a dropdown a CSM is expected to update once a quarter, and no system has been able to answer that at scale with the right object.
AI is rewriting the rules of what's possible, and the rewriting cuts across all three. For the first time, real-time feedback loops on value delivery actually can exist. The structured story of customer value realization is finally capturable from the conversations your team is already having. And the question of whether work is getting done the right way finally has an answer.
The winners of this era won't be the companies that bolt AI onto existing infrastructure to run old workflows faster. They'll be the companies that use AI to build a new layer on top of everything — a continuous, structured understanding of how they operate, how they deliver, and how every customer is doing — and use that layer to drive better customer outcomes more efficiently.
We call that layer Value Intelligence. We built Foresight to produce it in real time and drive the right action on top of it.
Why this matters now
The pressure on Customer Success teams has been mounting for years, and it's about to get worse.
Cassie Young at Primary Venture Partners has been warning about a gross retention apocalypse — the coming reckoning for AI-native companies that grew on experimental revenue without ever proving durable customer value. Her argument: ARR alone doesn't cut it anymore. The companies that survive will be the ones that "treat customer success not as a function, but as a core operating philosophy."
She's putting into language what CS leaders have been feeling for years: despite a decade of investment in CS tooling, the available stack has never been able to close the gap.
Health scores measure activity, not value. CRMs hold the data your reps wrote down, not the data they observed. Conversation intelligence captures what was said but not what it means for whether the customer is getting what they came for. None of these tools were designed around the question that actually matters in the AI era: is this customer realizing the value we promised — and if not, what do we do about it?
That's the gap Value Intelligence closes.
What four years of doing it by hand proved
I started Foresight five years ago. Before that I managed a $20M+ book of business as a post-sales leader, and my job — like every CS leader's job — was to keep customers and grow them. What was painfully apparent, even in a data-forward organization that believed in information, was the complete disconnect between the data we had and our customers' commercial decisions.
The data we had — and what our health score was derived from — was activity and user behavior: logins, NPS, support volume, stakeholders engaged. The data that mattered never showed up in a QBR. Without fail, executives didn't ask how often their team logged in. They asked whether their business was getting what they'd paid for, and if not, what we were going to do to deliver. They talked about outcomes and progress against the objectives that justified the purchase. That was the conversation that mattered, and far too often we had nothing structured to bring to it.
The insight that started Foresight was simple: the data set that matters in customer success isn't observations of user behavior. It's a structured understanding of what your customer was trying to achieve, how your business delivers value against it, and whether they're actually getting there. Not what they clicked, but what they came for and how it's going.
Every senior post-sales leader lights up when you ask them about this. They've all had the experience of it working — a customer whose outcomes you tracked so tightly that the expansion, the advocacy, the referral all followed inevitably. The problem was never that the industry didn't know this data mattered. The problem was that nobody could capture it at scale.
So we built it and brought what CS leaders had been managing in spreadsheets to life in a software platform: a value framework anchored in outcomes, each customer's experience measured against it, and team actions aligned to close the gaps. It worked, and customers we ran this way had the outcomes you'd expect — retention, expansion, and the advocacy that followed inevitably when CS work was being aimed at the right thing.
Then AI started rewriting the rules. Capability by capability, work that used to require human time, energy, and brain power became automatable, and what was once manual became more powerful in the process.
The customer value story is being told every day — in calls, in emails, in meetings, in the hundreds of small signals customer-facing teams pick up and forget. With the right capture architecture, AI now extracts those signals from every conversation, structures them against a model of how the business delivers value, and produces something that didn't exist before at scale: a continuous, complete, grounded read of every customer's value realization, captured automatically, every day.
That's what Value Intelligence is. Not a faster summary, a smarter dashboard, or a tagged "signal." A new layer of structured customer intelligence that answers the questions that actually matter — previously confined to the heads of your best operators, now captured at scale, across every customer, every call, every support ticket, every email, every day.
What gets built on top of it
A new layer of data is only valuable if something acts on it, and that's where most of the AI-for-CS conversation breaks down. Vendors talk about insight without ever closing the loop to action.
Foresight closes that loop, sitting on top of the Value Intelligence layer and doing the work no rep can do — and no rep should have to.
The architecture of that work is what makes the difference. The old way of operationalizing CS — the way every CSP and workflow tool does it today — is a static playbook. Someone in ops spends weeks configuring a sequence of triggers, alerts, and tasks in the CRM/CSP. The playbook is a best guess at what should happen, and once it's built, it rarely changes. The customer realities the playbook was meant to serve drift from the static script the moment it ships.
Foresight runs the playbook differently. The arc of the work — what good looks like for a customer at each stage of their value realization — lives in the system, but the execution is dynamic. What gets done for any given customer is shaped by what the Value Intelligence layer is showing about that customer right now, not by what an ops team guessed at six months ago. The playbook learns. The arc holds; the path adapts.
The arc holds; the path adapts.
That architecture is what makes the rest of the work possible. Watching every signal across every account. Holding your operating playbook in memory and making sure your team follows it. Preparing every team member for every customer interaction with the context, the history, and the right next move. Drafting the strategic email your best CSM would have spent Sunday night writing. Surfacing the patterns across your portfolio that no individual rep can see.
Foresight does that work, around the clock. A workhorse for your team, and a force multiplier for your company.
So every customer gets your best work — and your company grows what it delivers, not what it costs.
Don't buy the faster horse
It's unlikely Henry Ford actually said the "if you asked people what they wanted, they would have said faster horses" line, but the line itself has stuck in the cultural dictionary because of the truth it points to: it's difficult for people to reconcile what they've always known in the face of a transformative technology. The initial reaction is typically to do the thing I used to do, but faster.
The customer success industry is being sold faster horses right now — as documented in recent Bain & Co. research that 70% of CS leaders are stuck using AI in low level use cases: AI features welded onto architectures that were never designed to measure value, health scores reweighted with new math, dashboards refreshed with summarization, and off-the-shelf AI pointed at the data you already have, presented as a substitute for the underlying data layer. Every demo feels like magic for twenty minutes and a wall for the rest of the relationship.
The buyers who define the next decade of Customer Success won't ask for faster horses. They'll stop pretending the old architecture was measuring the right thing, and they'll demand systems built around what customers actually came for — with grounded action that delivers it.
That's the chapter we're writing. Not a better version of what we had, but a new layer on top of everything — finally answering the questions every executive has been asking for a decade.
What this looks like from inside the day-to-day, when an operator imagines it from the seat of a CS leader running a real team, is the next thing worth writing about. More on that soon.