You’d assume Anthropic, of all companies, would be running on some self-grown, cutting-edge AI-native sales platform. Maybe a Salesforce killer they built themselves. Maybe Claude baked into the protocol layer of every workflow.
They aren’t. At least not yet. Claude is baked into their stage, it’s the substrate. But it’s used to connect to the core B2B apps so many of us also run on.
When Eleanor Dorfman, Anthropic’s Head of Industries, walked through the actual GTM stack at SaaStr AI Annual 2026, the lineup was the most “boring” thing about this highly rated presentation. It’s the same six tools so many other leading B2B + AI companies uses, plus a familiar supporting cast of Jira, Intercom Fin, Snowflake, BigQuery, and G Suite.
The interesting part is what they do with each of them.
A guided tour.
Clay: Used to Be an Enrichment Tool. Now It’s the Funnel Gate.
Most teams: Use Clay to enrich inbound leads with firmographic data. A human reviews the enriched record and decides what to do next.
Anthropic: Clay + Claude qualify every lead at the moment of capture and decide which of two funnels the lead belongs in. The qualification step isn’t a human looking at enrichment data. It’s Claude evaluating fit signals and routing the lead to either the AE-led path or the self-serve path within seconds of capture.
The result: 54% of new enterprise logos in 2026 came through the self-serve path. The Clay + Claude qualification layer is what made that gate possible.
LeanData: Used to Be a Router. Now It Routes Between Humans and AI.
Most teams: Use LeanData to route qualified leads to the right AE based on territory, segment, or product line.
Anthropic: LeanData still routes, but the destinations now include not just AEs but AI funnels. Leads can route to a BDR queue, an AE directly, or to the Intercom Fin guided self-serve flow. The router is no longer just deciding which human gets the lead. It’s deciding whether a human gets it at all.
Salesforce: Used to Be Where AEs Worked. Now It’s Where Claude Updates.
Most teams: AEs manually log calls, update opportunity records, set next steps, and write account notes in Salesforce. The system of record runs on rep discipline.
Anthropic: Claude updates Salesforce. The morning briefing reconciles opportunity records against context pulled from Gong calls, emails, and Slack threads. AEs inspect and approve. The forecast call is no longer a data-scrubbing exercise where managers verify Salesforce hygiene. It’s a discussion about where AEs need help, because the data is already current.
A subtle shift with big implications. Salesforce stops being the place reps go to enter data and becomes the place Claude goes to reconcile state.
Gong: Used to Be for Call Coaching. Now It’s a Context Source for Everything.
Most teams: Sales managers listen to Gong calls and deliver coaching based on a methodology. Reps revisit recordings to remember what was said.
Anthropic: Gong is the highest-value context source in the entire stack. Claude pulls Gong transcripts into the morning briefing, the call prep Skill, the proposal-drafting Skill, and the weekly coaching loop. When a rep types /call prep before a meeting, Claude is reading Gong transcripts of every prior call with that account to generate the briefing. The coaching loop itself is now dynamic. Claude surfaces six coaching moments per week that change based on what matters this month, not what a static methodology says matters this year.
Gong didn’t change. The number of places its data flows into multiplied by ten.
Ironclad: Used to Be Where Contracts Got Signed. Now It’s Where Claude Ships Proposals.
Most teams: AEs draft proposals manually. Open nine tabs of deal desk guidance. Scrub Gong transcripts to remember what was negotiated. Upload the proposal to Ironclad for redline cycles.
Anthropic: Claude does the drafting. With one prompt, Claude pulls customer history from Salesforce, negotiation context from Gong, policy guidance from internal docs, and precedent from prior deals. It drafts the proposal within policy, uploads it to Ironclad, and kicks off the workflow.
Eleanor’s framing: instead of an AE opening nine tabs and listening to Gong calls to reconstruct what had been negotiated, Claude generates the proposal and uploads it to Ironclad.
The contracting tool didn’t change. The work happening before the contract gets there collapsed from hours to one prompt.
Slack: Used to Be Where Deals Got Coordinated. Now It’s the Front Door for Everything.
Most teams: Slack is where AEs coordinate with deal desk, legal, RevOps, and customer support. Often through DMs. Often based on which AE happens to know which person on which floor.
Anthropic: Slack is the single front door for every support function. AEs file a ticket in Slack. Claude triages. If the question matches precedent, Claude resolves it inline. If not, Claude attaches the full context (Salesforce history, Gong summaries, email threads, customer contacts) and routes to a human, generating a Jira ticket on the back end.
Slack ticket in. Jira ticket out. The DM-based system that ran on institutional knowledge is gone.
Intercom Fin: Used to Be a Support Product. Now It Closes Enterprise Deals.
The most counterintuitive move on the list.
Most teams: Intercom Fin is the customer support deflection tool. It handles support tickets, answers FAQs, escalates to humans when needed.
Anthropic: Fin is the seller for the enterprise self-serve funnel. After Clay + Claude qualify a lead into the self-serve path, Fin takes over and guides the buyer through enterprise plan selection, terms of service, invoicing, provisioning, and onboarding enrollment. Anthropic partnered with the Intercom Fin team to retool a flagship support product into a viable enterprise sales tool.
A product built to answer “how do I reset my password” now closes enterprise contracts with real ACV. If you’re wondering what Intercom’s stock is going to do in 2026, this is part of the answer.
Jira: Used to Be for Engineering Tickets. Now It’s the Sales Support Backbone.
Most teams: Jira is where engineering tracks tickets and bugs. Sales rarely touches it.
Anthropic: Jira is the back-end ticketing system that catches escalations from the Slack front door. Sales-adjacent functions (deal desk, legal, vendor onboarding, security review) run their workflow in Jira. Claude generates the tickets with full context, routes them, and tracks status back to the AE in Slack.
Jira wasn’t designed for sales coordination. It just happens to be the cheapest, most flexible workflow tool already deployed across the company. Rewiring an existing tool is faster than buying a new one and training the team on it.
Snowflake and BigQuery: Used to Be for Analytics. Now They Power Forecasts.
Most teams: BI tools to track sales metrics, generate dashboards for QBRs, and report up the chain.
Anthropic: Snowflake and BigQuery power Claude’s forecasting motion. Anthropic is a consumption business, so Claude reads actual usage patterns from the data warehouse and reconciles them against historical patterns for each customer cohort and product mix. The forecast call is no longer “let’s project from this quarter’s commits.” It’s “Claude has analyzed actual consumption and identified the deviations from expected. Let’s discuss those.”
Forecasting moved from a guess-the-spread exercise to a consumption-driven inference. The data warehouse is the substrate.
G Suite and Greenhouse: Used to Be Office Software. Now They’re Context Surfaces.
Most teams: Gmail, Google Docs, Calendar are where work happens. Greenhouse is where recruiting happens. None of them feed into the sales motion meaningfully.
Anthropic: Every one of these is a connector into Claude. The morning briefing reads your calendar, your email, your Slack, your Salesforce, your Intercom, and your Greenhouse simultaneously to prioritize the day. The customer follow-up Skill extracts action items from Gmail and drafts responses. Greenhouse data feeds account intelligence: a customer hiring 50 engineers in the next quarter is a different sales signal than one that’s flat on headcount.
Office software stopped being office software. It became a context graph.
Claude: Not a Seventh Tool. The Substrate.
The trap most B2B + AI sales leaders fall into is treating Claude (or any AI) as another tool to add to the stack. Buy a license, set up the integration, ship it to reps, watch them not use it.
Anthropic didn’t do that. Claude isn’t a seventh tool. Claude is the substrate every other tool sits on. It reads from all of them, writes to all of them, and operates as the orchestration layer between them. Reps don’t open Claude as an app to use. They summon Claude with a / shortcut from wherever they’re already working.
The five Skills are how reps interact with the substrate: morning brief, call prep, customer follow-up, competitive intel, and create-an-asset, plus the proposal drafting workflow that happens through Ironclad. Each Skill pulls from multiple connectors and writes back to whichever tool the work belongs in.
A Few More Notes from Eleanor Worth Knowing
- Anthropic AEs are paid very well. OTE ranges from $270K to $445K, with top performers earning $710K to $1.56M. 87% of AEs hit their number. The most efficient sales org ever built also has some of the highest-paid AEs in the industry. The role isn’t getting eliminated. It’s getting leveraged, and the comp reflects it.
- 87% hit their number on shadow targets. Anthropic doesn’t run hard quotas. They run shadow targets, then pay people like they hit hard quotas. The accountability lives in the workflow design, not in the comp plan structure.
- Skills are summoned with
/shortcuts from inside the tool where work is already happening. Reps don’t context-switch to a “Claude app.” They type/call prep,/morning brief,/follow up,/competitive intel, or/create assetfrom Slack, Salesforce, or wherever they are. The interface is invisible. - The 24-hour customer follow-up SLA is enforced by Skill, not by manager. The customer follow-up Skill drafts the responses, drops them in your email, and reminds you in tomorrow’s morning brief if you didn’t ship them. Accountability moved from 1:1s to the workflow itself.
- Co-work updated the playbook in real-time. Eleanor noted that Co-work changed how the email-to-Slack-ticket triage works in the week before her SaaStr talk. The stack isn’t static. Tools change weekly. The companies winning are the ones with the architectural flexibility to absorb that pace.
- “AI slop” is the failure mode Anthropic actively engineers against. Skills are designed to know the brand, the policy, the customer context. Reps are explicitly trained to never ship AI slop to customers. Brand integrity is a first-class design concern in the Skills layer, not an afterthought.
- Eleanor’s line worth tattooing: “Sales leaders are becoming systems thinkers over deal strategists.” The job changed. You’re now designing the system that produces deal velocity, not the deal itself. The leaders who can’t think in this mode will struggle to keep up.
The Stack You Already Have Is Probably Fine. It’s Time To Use It Much More Effectively With AI.
The lesson here isn’t that Anthropic discovered a magic stack. They didn’t. They use the same names you do. Clay. LeanData. Salesforce. Gong. Ironclad. Slack. Jira. Intercom Fin. Snowflake. BigQuery. G Suite.
What changed isn’t the names. None of those tools are end destinations anymore. Each one is a surface that Claude reads from, writes to, and connects to the others.
You don’t need a new CRM, a new call coaching tool, or an AI-first sales platform. What you need is to thread an AI substrate through the stack you’ve already paid for, trained on, and built your motion around.
The most efficient sales org ever built runs on the same software you do. It just doesn’t need any much of it as many do.
