How many employees can I replace with AI?

Bar Nakash — Wed, 28 Oct 2026 11:44:28 +0000

Originally published by Bar Nakash on LinkedIn (July 15, as displayed in the source).

This sentence is everywhere right now, and I get it. As a CEO one of the core responsibilities is making a company more profitable, and when a tool comes along that can do in minutes what used to take hours, the math is tempting.

I think we're asking the wrong question, it shouldn't be who can I replace with AI, but how can AI make the team I already have even better.

At Applift, every single person on the team uses AI daily, and we're not ashamed of it. It makes us faster, it spots mistakes we missed, it gives us ideas when the creative well runs dry, and it automates the repeating tasks that used to take an entire day out of our work week. I really can't remember how we used to work without it.

And still, our real achievements were made by our team. The pitch that won our biggest account this year, the insight that turned a struggling campaign into a case study, the call at 10pm on a Thursday because a client needed someone who actually understood their business. All of that was done by a person.

Every person at the Applift team was chosen because they're exceptional at what they do. AI is the best tool we've ever given them. But a tool doesn't think, and they do.

AI is the sharpest tool this industry has ever seen. But a tool without the right person behind it is just expensive software.

So when someone asks me how many employees I can replace with AI, I have an answer. None. I'd rather give exceptional people an unfair advantage than hand the work to a machine that doesn't know why it matters.

Before You Change an App Listing, Write a Measurement Contract

Bar Nakash — Tue, 06 Oct 2026 18:25:42 +0000

A store listing test is easier to interpret when the team decides what success means before changing the creative. Here is a small measurement contract an app team can keep beside its release notes.

1. State one hypothesis

Showing the core workflow in the first screenshot will make the benefit clearer than an abstract headline. This is a hypothetical test, not a reported client result. Identify the element you will change and the audience it is meant to help.

2. Name the report and metric

Record the store, reporting screen, metric label, definition, date range, territory, language, and user segment. Keep a baseline export. Avoid combining two charts just because both use a word such as conversion: their denominators and populations may differ.

3. Keep a change log

Listing edits can coincide with app releases, campaigns, pricing changes, publicity, and outages. Log those events with dates. A before-and-after increase is an observation; it does not establish that the listing change caused all of it.

4. Match the test to the question

Google Play supports experiments for listing graphics and text, with results tied to the chosen target metric. Its documentation recommends changing one asset at a time and explains confidence intervals and minimum detectable effects. Check the current console rather than assuming the test setup stayed the same.

A screenshot experiment addresses the response to that tested creative among the exposed audience. It does not automatically answer whether search visibility, retention, or revenue improved.

5. Write the decision rule

Use a short template:

Hypothesis and audience:
Control and changed element:
Primary metric and exact definition:
Product guardrail:
Concurrent changes:
Evidence needed for rollout:
Result and remaining uncertainty:

Treat a draw or an insufficient-data result as a reason to inspect the evidence, not an invitation to invent a winning narrative. If a controlled test is unavailable, describe the comparison as directional and record its limits.

6. Connect the listing to the app

A clearer listing should set an expectation the product can deliver. Review activation and retention separately from store-level outcomes. If the data cannot be reliably joined, report the two views separately.

The useful output is a decision: keep the current listing, roll out a supported treatment, or investigate another hypothesis.

Source: Google Play Console: Run A/B tests on your store listing, checked October 6, 2026. The measurement-contract framework above is practical guidance, not a Google ranking formula.

DEV Community: Bar Nakash

How many employees can I replace with AI?

Before You Change an App Listing, Write a Measurement Contract

1. State one hypothesis

2. Name the report and metric

3. Keep a change log

4. Match the test to the question

5. Write the decision rule

6. Connect the listing to the app