/Tech3h ago

11x CEO Details Internal AI Agent Stack in a16z Interview

1772184521K

#524

Original post

Dani Grant@thedanigrant

I'm talking to @a16z founders about how they use AI backstage.

Here's 11x's internal AI operating stack (they use to build AI agents):

a16z@a16z

.@11x_official's revenue agents are generating hundreds of millions of dollars in pipeline for their customers.

Here's their full stack of internal agents, AI they run inside their own company, so that they can spend maximum time building agents for customers.

We sat down with CEO @prabhavjain and went through their full internal AI operating stack:

10:31 AM · Jun 25, 2026 · 3.5K Views

Sentiment

Users praise 11x's internal AI agent stack as impressive because it prioritizes in-house efficiency optimization as a key differentiator.

Pos

100.0%

Neg

0.0%

2 comments with sentiment.

Cluster Engagement

Digg Deeper

No Digg Deeper questions have been answered for this story yet.

Posts from X

Most Activity

VIEWS1.6K

a16z@a16z

One pattern we're tracking: the technical CEO.

@prabhavjain led eng at @brexHQ & was @11x_official's CTO before he was CEO.

A company that runs on agents is a system design problem.

And that's creating a lot of horsepower for engineers in the CEO seat.

Watching this space.

3h1.6K11

BOOKMARKS2

a16z@a16z

PS – 11x is hiring: • sales • success • growth • engineering • design

https://jobs.a16z.com/jobs/11x.ai

3h1.6K12

LIKES2

Dani Grant@thedanigrant

We're posting a full write-up of their internal stack tomorrow on a16z Build, watch this space 😃:

http://a16zbuild.substack.com

3h572

RETWEETS2

a16z@a16z

.@11x_official's revenue agents are generating hundreds of millions of dollars in pipeline for their customers.

Here's their full stack of internal agents, AI they run inside their own company, so that they can spend maximum time building agents for customers.

We sat down with CEO @prabhavjain and went through their full internal AI operating stack:

3h17.5K6337

REPLIES1

Bright Chukwuemeka@BrightChuk2sf0

@a16z @11x_official 220 likes @tweem

3h1

a16z@a16z

1/ Customer qualification skill

When @prabhavjain took over from CTO to CEO, he made a rule: They are only selling to customers who will see incredible value from 11x. So he built a Claude skill every sales rep uses.

It pulls in emails, call notes, Slack messages exchanged with prospects. It considers industry-specific nuances and patterns from past successful customers.

If it's missing information, it tells the rep what questions to ask. The agent gives the rep a confidence score whether the customer is a fit or not. Reps only close deals with confidence >85%.

3h3951

a16z@a16z

2/ Deal handoff agent

Customer satisfaction = expectations (what was promised during sales) minus reality (what the account team helps deliver)

To make sure every detail of what was promised upfront is communicated to the post-sale team, they built an agent.

It runs on custom code and you can assign it to grade a deal in Slack by @ mentioning it.

Claude integrations pull every call from @meetgranola, Slack, and email exchanged across the whole cycle.

Then it generates a Notion doc and a kickoff deck.

Agents are good at context, humans are good at service.

3h3101

a16z@a16z

Stack at a glance:

• Reasoning / agent layer: @claudeai Cowork, Claude Skills • Orchestration: @LangChain Fleet + @SlackHQ bots • Meetings: @meetgranola • Docs/decks: @NotionHQ • Code / Code harness: @claudeai code + @cursor_ai + @coderhq

3h1911

a16z@a16z

One thing to really like about this stack is it's simple; it's not over-engineered.

Outside of the code harness, it's mostly Claude Cowork + MCPs running at timed intervals. They are getting value by having good ideas of what to automate, not by having fancier tools.

The framework: basic, boring tech that just works, doesn't break, and you don't have to fuss with.

3h1871

a16z@a16z

5/ The agentic office of the CEO

CEO @prabhavjain built a simple fleet of personal agents with intel for each day, all running on Claude Cowork scheduled jobs.

The first is a customer brief, delivered via Slack DM each week. It ingests every customer call via the @meetgranola MCP. This gives him a weekly update on what matters for customers right now.

The second is a meeting prep agent that reads email, Slack, Granola, conducts research and preps him for the next day's meetings.

The third combs through his Slack every evening, and flags unresolved threads. Resolution isn't always explicit, it could be just an emoji reaction, or it could be in a different DM than the main discussion thread, so the agent has to reason.

A fourth is a weekly market pulse. It pulls from the internet, and also from objections and competitors mentioned in that week's deals in calls and emails. This gives him a pulse on what the field is seeing.

3h1731

a16z@a16z

3/ "Lemmings"

11x runs a company-wide fleet of internal codebase-read agents they call lemmings. The non-engineering team dispatches lemmings via Slack to answer questions and build a handful of customer-facing docs:

• Create customer docs based on codebase, tailored to specific customer • Generate insights for a customer, and put into a polished doc • Debug and optimize a customer's setup • Answer customer questions based on product and outcome data • Create decks for quarterly customer reviews

This is built with @LangChain's LangSmith Fleet, which lets you describe in plain english a fleet of agents that work across your internal code and tools, and at mention them from Slack.

3h1701

a16z@a16z

4/ Coding agent + test/review harness

Like all startups, 11x uses coding agents heavily.

But 11x also realized that as coding agents wrote more PRs, the bottleneck would move to verifying PRs.

So early on, they decided to invest heavily in building their own code harness.

The code harness verifies PRs created by agents by: • Loading up a full virtual environment w/ @coderhq + @Docker • Figures out what the PR is about and makes a test plan • Runs e2e automated tests in the virtual environment via @playwrightweb • Runs existing suite of regression tests • Puts evidence of the new feature working via screenshot in PR • Has mock data that mimics production (including volume) and ensures performance efficiency doesn’t drop with any new PR

3h1611

coc@mrcoc0x

@a16z @11x_official that's really impressive 🤩, would love to learn more about their strategy 💡

3h131

Crypto Briefing@Crypto_Briefing

@thedanigrant @a16z

3h17

The White Dolphin@WhiteDolphinSol

@a16z @11x_official We should all buy Bitcoin it's cheap rn

3h5

tweem@tweem

@BrightChuk2sf0 @a16z @11x_official

3h2

Emmanuel Ohajuru(Noya)π²🌶️@OhajuruEmmanuel

@a16z @11x_official @tweem 4k likes

3h2

Adel Bucetta@adelbucetta

@a16z @11x_official they're optimizing for in-house efficiency, not just customer-facing revenue. that's what sets apart companies who truly master ai integration from those who just automate tasks.