Here is a thing that is true about enterprise software sales, and that everyone in enterprise software sales quietly knows: most of the job is not selling. Most of the job is preparing to sell. It is reading the 10-K before the call. It is scrolling three years of CRM notes to remember who got promoted and who got fired. It is listening back to a Gong recording at 1.5x to figure out what the buyer actually cares about, as opposed to what the buyer said they cared about, which are frequently different things.
This is unglamorous, and it does not scale, and it is exactly the part of the job that AI turns out to be pretty good at. So the pitch for Endgame, a San Francisco company founded in 2021 by Alex Bilmes, is roughly: what if a machine did the homework? Not the sending-emails part - there are, God knows, enough tools that will send more emails - but the knowing part. The part where a seller shows up to a meeting having read everything, remembered everything, and connected it to what the buyer's CFO said on the last earnings call.
Endgame describes what it is building as a "knowledge system" for enterprise sales, and, more recently, as "the context graph for every GTM agent." Both phrases are doing a lot of work, and we'll get to the second one, because it is the more interesting claim. But the plain-English version is this: connect the company's CRM, its email, and its call recorder; blend that first-party data with the outside world - news, financial filings, LinkedIn; and then let a seller do account research, meeting prep, deal inspection, and quarterly business reviews roughly, per the company, 100x faster.
"100x faster" is the kind of number that should make you narrow your eyes, and you should. But the underlying observation is sound. If the bottleneck in a job is preparation, and preparation is mostly retrieval and synthesis, then a tool that is very good at retrieval and synthesis is genuinely 100x on that specific task, even if it's 1x on the parts that involve a human being liking another human being.
The category it named, and outgrew
There's a nice arc here. When Endgame started, its idea was "product-led sales," or PLS, a term the company helped popularize. The theory of PLS is that in modern software - the kind you can just sign up for and start using - the product itself generates the best sales signals. Somebody at a big company spins up a free Figma workspace, invites forty coworkers, and starts using it every day. That is a better buying signal than any form they'll ever fill out. PLS software watched product usage, plus the CRM and the data warehouse, and told sales teams which accounts and users to call and why.
It worked well enough that in 2021 Endgame raised more than $17 million, and in February 2022 it closed a $30 million Series B led by EQT Ventures - a round the company has said it wasn't even planning to raise, but took because early design partners started paying faster than expected. That is, as fundraising stories go, a good problem. Total funding sits at roughly $47.5 million, from a cap table that includes Menlo Ventures, Upfront Ventures, Unusual Ventures, and operator-angels from Zoom, Airtable, Stripe, and Notion.
And then generative AI happened, and the ground moved. To Endgame's credit, it moved too. In November 2024 it shipped Endgame 2.0, an AI-native rebuild aimed squarely at enterprise sellers. The company is fairly explicit that this is a different product from the one it raised on, which is the correct and slightly uncomfortable thing to be explicit about. Naming a category is a great way to win the last war. Endgame decided it would rather win the next one.
What makes Endgame 2.0 more than a chatbot bolted onto a CRM is that the company insists it is opinionated. It structures information around how enterprise sales is actually supposed to work - what questions to ask, what risks to flag, what use cases to chase - rather than just answering whatever you type. That's a real product stance. A generic AI will retrieve. An opinionated one will tell you that you forgot to ask about the renewal date, and here's why that matters.
The interesting, slightly scary bet
Now, the "context graph for every GTM agent" line. The bet buried in it is that as every revenue team spins up its own fleet of AI agents - one to research, one to draft, one to forecast - the scarce resource stops being the model. Anyone can call a model. The scarce resource becomes the shared, trustworthy context those models draw from: the assembled, deduplicated, source-cited understanding of who this account is and what's true about it. Endgame wants to be that layer. Not the agent - the thing the agents ask.
If that's right, it's a much bigger business than "a nice research tool for sellers," which is presumably why the company phrases it that way. It's also harder, because being infrastructure means other people's products break when you're wrong, and Endgame writes candidly about that - about "the prototype gap," the way an AI demo can dazzle in five minutes and quietly fall apart in production. Their public field notes go deep on the boring, load-bearing layers: retrieval, provenance, evaluation, observability. The demo, they argue, is the easy 10%. The other 90% is why most AI products never actually ship. It is refreshing to hear a startup say the boring part out loud.
The customers, the competition, and the small-team math
The customer list is itself an argument. Over its life Endgame has been used by, or reported alongside, a roster that reads like a tour of modern software: Figma, Calendly, Loom, LaunchDarkly, Airbyte, Retool, Algolia, and Scale AI. There's a logic to that. If you want to sell a new idea about how software gets sold, you sell it first to the companies already living the problem - the ones whose products spread bottom-up through organizations and whose sales teams are drowning in usage signals they don't have time to read. Those customers become your proof, your feedback loop, and your case studies all at once, which is an efficient way to run a young company.
The competitive picture is more crowded than it was in 2021, because everyone now has an "AI for sales" story. Endgame sits in a neighborhood that includes revenue-intelligence incumbents like Gong and Clari, activity-capture players like People.ai, and a fresh crop of AI account-research and sales-copilot startups, plus the ever-present option of a company's own engineers stitching something together with off-the-shelf models. Endgame's answer to all of it is the same: the model is a commodity, the context is not. What's hard is assembling a trustworthy, source-cited understanding of an account and keeping it current. That's the moat it's trying to dig, and it's a defensible place to dig one.
There's also a headcount detail worth sitting with. Endgame's own about page has described a team of around 21 people, though third-party data services list closer to 45. Either way, this is not a large company, and that's part of the thesis rather than a footnote to it. The whole bet - "help humans think and work better" - is that a small, senior team armed with the right AI can out-research a sales organization many times its size. If that's true for Endgame's own operation, it's presumably true for its customers, which is convenient, because it means the company is its own best demo. The angel list - operators from Zoom, Airtable, Stripe, and Notion - suggests the people who've built go-to-market machines at scale found the pitch credible enough to write checks.
So what can you actually do with it? If you run or work on an enterprise sales team - especially one living in Salesforce and Gong, which Endgame prioritizes - you point it at an account and ask anything. It reads the whole relationship plus the outside world and hands back a cited answer, in Slack, in email, wherever you work. The promise is not that it closes the deal. The promise is that the seller who uses it is the best-prepared person in the room, every time, without having spent the morning preparing. Whether that makes them a "trusted advisor" or just a very well-briefed one is, pleasantly, a distinction the buyer never has to notice.