Knock · Marketing Engineer Meetup · Sept 15, 2026

Agentic Onboarding

What changes when your new users are agents

Tommy Keeley
Head of Growth at Amplitude
Sept 15, 2026
Amplitude

Who am I

Tommy Keeley is Head of Growth at Amplitude, where his team drives AI adoption & builds agentic capabilities into products like Slack & Teams. Before Amplitude he spent four years driving growth at Dropbox, including Principal PM for Sharing & Virality & interim Head of Growth. Prior to Dropbox, Tommy worked at high growth startups including Attentive & AlphaSense.

Amplitude Dropbox Attentive AlphaSense

Agenda

  1. 1The Basicsonboarding 101
  2. 2Agentic Acquisitionfind › pick › signup
  3. 3Agentic Activationsetup › aha › habit
  4. 4Demo
  5. 5Conclusionwhat means success

Part 1

The Basics

Introduction

The Growth Funnel

Sources
SEO SEM Social Blogs Word of mouth Referral Partnerships Direct
1

Acquisition

Discover and enter your product

MetricSignups
2

Activation

Experience value for the first time

Setup↓ Aha↓ Habit
Metric7D Aha Rate
3

Engagement

Keep coming back for value

MetricWeekly Active Users
4

Monetization

Turn product value into revenue

MetricTrial to Paid CVR

Part 2

Agentic Acquisition

  1. 1Find
  2. 2Pick
  3. 3Signup

Agentic Acquisition

Find › Pick › Signup

Your documentation is your homepage now
The question used to be "can a person find our website?"

agent
$

Can Agents find your product?

  1. What it already knowsOften the agent never searches. It picks what it remembers from training, or what is already installed in the repo.
  2. Web search rank and citationsRank for the task on Bing and Brave, not just Google.BingBrave
  3. Docs, READMEs, registries, reposPlain text, listed where it looks.GitHub
  4. Robots and rate limitsCrawlable is the new indexed. If GPTBot and ClaudeBot are blocked or throttled, the agent cannot read you, so you do not exist.AI agent
Mintlify

45% of docs requests now come from agents, not humans.

Agentic Acquisition

Find › Pick › Signup

Find puts you in the set. Pick is whether you win the comparison.
Like a human, an agent compares fit, reads reviews, & explores evidence, deciding which product is right for the user & task?

Does your product get picked?

  1. Fits the task.Show the jobs you solve with clear examples.
  2. Fits the user.Be clear who your product is for.
  3. Earns trust.Build trust through reviews and customer proof.
  4. Backs up the promise.Prove it with results, current docs, and a working trial.
Mixpanel
  • Fits the task
  • Fits the user
  • Strong reviews
  • Proven results
Picked
Amplitude
  • Fits the task
  • Fits the user
  • Strong reviews
  • Proven results

Best fit for this task

Google Analytics
  • Fits the task
  • Fits the user
  • Strong reviews
  • Proven results

Part 3

Agentic Activation

  1. 1Setup
  2. 2Aha
  3. 3Habit

Agentic Activation

Setup › Aha › Habit

Setup prepares an agent to get value from your product.
For Amplitude: install the SDK, send the first events, and verify they arrived.

Can an agent set itself up?

  1. Instructions it can follow.Follow one versioned set of install and event-tracking steps.
  2. The key & the project from signup.Pass the key and project directly from signup.
  3. A way to check it worked.Confirm events arrived with a count from one API call.
Install the SDKnpm i @amplitude/analytics-browserInstrument eventstrack('Checkout Started')Verify they arrived✓ 12 events received Set up! First event in the product

Agentic Activation

Setup › Aha › Habit

The aha is the first meaningful thing done with the data.
Build the first chart or dashboard, answer a real question, and verify the result.

Does the agent reach the aha?

  1. Do something with it.Use the data to answer a real question.
  2. Do it first try.Make the first meaningful action succeed.
  3. Check it's real value.Verify the result matches the data and the user’s question.
Events landed12 events receivedBuild the first chartCheckout funnel, last 7 daysAnswer a questionDrop-off at payment step: 38% Aha! Checked against the events it sent

Agentic Activation

Setup › Aha › Habit

The benefit of an agent is that it never forgets.
Turn the first win into a recurring job. Verify it and schedule the next run.

Does the agent come back?

  1. Leave a hook.Offer a recurring digest, alert, or check after the first win.
  2. Make it easy to repeat.Provide a reusable skill or plugin with clear examples.
  3. Give it a reason to return.Save a job tied to the user’s goal. Verify it and schedule the next run.
Leave a hookWeekly digest of the checkout funnelSave the jobMondays at 9, results to SlackVerify the first runFunnel checked, report delivered Habit! Runs every Monday

Part 4

Demo

Demo

One starting point. One setup flow.

The same prompt for a human or an agent.

One path from entry points to initial setup Human and Agent converge on The Entry Points, which lead to The Prompt, the single starting point. The Prompt connects to Amp CLI for deterministic actions. The yellow Entry Points bar connects to the yellow Prompt card. The blue Prompt bar connects to the blue CLI card. The green CLI bar connects to the green Skill and Tools card, whose blue bar connects back to the CLI. A dashed line on the right marks initial setup complete. THE ACTOR Human Agent The Entry Points Many doors. One path. SINGLE STARTING POINT The Prompt One way to get started. Amp CLI Deterministic actions. >_ REMOTE BRAIN Skill & Tools Reusable instructions. Initialsetupcomplete

Demo

Agentic setup walkthrough

Conclusion

What you can do for your product

Ask an agent to implement your product.
Turn what stalls into your roadmap. Then measure whether your improvements work.

Start with a one-hour audit

  1. Run it end to end.Ask an agent to sign up for your product and start using it.
    Do not help.
  2. Vary the attempt.Try a different model, harness, and way of asking.
  3. Ask what happened.What did it try? Where did it look? What would have made it pick you?
  4. Build your improvement list.Write down every stall, guess, and quiet wrong turn. That is your onboarding roadmap.

Measure as you improve

Find
Agent share of docs trafficAre agents finding and reading your documentation?
Pick
Pick rate by user and use caseAre you winning the comparisons that matter?
Signup
Hands-off completion rateCan the agent finish with zero human steps?
Activate
Time to first trusted outcome + verification rateDid it work, and can the agent prove it?

Repeat the same tasks after each change.

Thank you

Tomorrow, free on Maven

Win with Growth Loops in an Agentic World

Wed Sept 16 · 1:00pm ET · 45 min · Zoom

Scan code & save a seat!

QR code for the Maven lightning lesson
 X Find me on X @tommydotkeeley

Appendix

A quarter of setup in one session

Point an agent at a scratch repo. Watch it find the path, run the skill, verify, and hand a human a dashboard. Messy legacy codebases still need people. The interesting part is where the work moved.

agent · scratch-repo
$ press Run, or Enter
01Find the instructionWaiting
02Run the skillWaiting
03Verify with a validation IDWaiting
04Hand a human a dashboardWaiting

Simulated session for rehearsal. On the night this panel is the cue to switch to the live screen.