Imaguru x Andrew Tomin x axwise

Imaguru
AI Product Sprint

From customer evidence to a testable hypothesis and working prototype.
Monday, 27 July 2026 / 17:00-21:00 CEST / Imaguru, Warsaw
Vitalijs Visnevskis
AI Product & Technology Expert
Co-Founder, axwise.de
Andrew Tomin: AI Sandbox
Senior Product Manager, Agentic Ops
AI Trainer, aisandboxlab.com
1
Find a partner.
You'll work in twos.
2
Open the prompt pack:
aisandboxlab.com/imaguru-sprint
3
Sign in to your builder and publish a test project.
QR code to the prompt pack
Prompt pack
01
Agenda
17:30-21:00

Tonight's arc

17:30
Kickoff
10 min
17:40
Evidence-based discovery
25 min
18:05
Assumption & test design
25 min
18:30
Break
10 min
18:40
PRD-lite
30 min
19:10
Build one value loop
60 min
20:10
Publish & demo prep
20 min
20:30
Lightning demos & Q&A
30 min
02
Kickoff
17:30-17:40

Your trainers tonight

Vitalijs Visnevskis
AI Product & Technology Expert, Co-Founder of axwise.de
Andrew Tomin
Senior Product Manager, Agentic Ops, AI Trainer, AISandboxlab.com
03
Kickoff
17:30-17:40

What tonight actually is

Core promise

Leave with an evidence-informed problem hypothesis, a real-world validation plan, a one-page product brief, and a working prototype of one core value loop.

Important

This sprint launches validation, it does not validate a market in three hours. AI personas may challenge your assumptions. They are never customer evidence.

04
Kickoff
Definition of done

By 20:30 every team has

01
Evidence log
3+ sources, evidence separated from assumptions
02
Problem hypothesis
User, context, current workaround, biggest risk
03
Validation kit
5 interview questions, outreach message, decision signal
04
PRD-lite
One journey, acceptance tests, mock data, out of scope
05
Working prototype
One value loop, published, tested outside the editor
05
Kickoff
17:30-17:40

Pick your lane, know your limits

Beginner / default

Bolt, Lovable, or Replit Agent

No install. This is the right lane tonight.

Advanced / optional

Claude Code, Codex, Cursor

Or your own coding agent. Self-supported.

Build constraints
No billing
No custom auth
No complex infrastructure
One core feature
Every prompt you need tonight: aisandboxlab.com/imaguru-sprint
Open it now. You'll use it in five minutes.
QR code to the prompt pack
Prompt pack
06
25 min / 17:40-18:05
Block 01

Evidence-based discovery

Find real evidence your problem exists, before you build anything.
07
Evidence-based discovery
17:40-18:05

How to research it

Prompt pattern
"Research this problem as a skeptical product analyst. Cite sources, distinguish facts from inference, show current alternatives and workarounds, and identify the three riskiest unanswered questions."
Label every finding
Primary evidence
Secondary evidence
Inference
Output 1

One-sentence problem statement: user, trigger, problem, what they do today

Output 2

Evidence log with 3+ sources you can open

Output 3

The 3 riskiest unanswered questions

08
25 min / 18:05-18:30
Block 02

Assumption & test design

Find the one assumption that could kill this, and write how you'd really test it.
09
Assumption & test design
18:05-18:30

Build your validation kit

  • 1
    Rank assumptions: highest impact, weakest evidence wins
  • 2
    5 interview questions about past behaviornot opinions of your idea
  • 3
    1 outreach message to a real person matching your target profile
  • 4
    1 decision signal: continue, change direction, or stop
Reminder

Synthetic AI personas stay labeled "unverified". Never call them interviews, participants, or validation.

10
Break
18:30-18:40
10:00

Back at 18:40

Access issues or an oversized idea: flag a trainer now, not during the build block.

11
30 min / 18:40-19:10
Block 03

PRD-lite

One page, ten fields, small enough to build in under 60 minutes.
12
PRD-lite
18:40-19:10

The brief

01
Target user
02
Problem
03
Evidence
04
Assumptions
05
Core promise
06
One journey
07
Acceptance tests
08
Mock data
09
Out of scope
10
Next experiment
13
Guardrails
Before you build

Before you build: guardrails

Scope

One user, one value loop. Mock billing, auth, and integrations. No multi-tenant, no Docker, no dashboards.

Data & security

Never paste confidential data or API keys. Assume a published prototype is public until you've tested otherwise.

14
60 min / 19:10-20:10
Block 04

Build one value loop

One working, end-to-end interaction. Not a full product.
15
Build one value loop
19:10-20:10

Build discipline

  • 1
    Scaffold only the primary journey's screens
  • 2
    Mock data for everything that isn't the core interaction
  • 3
    Run the flow against your acceptance tests
  • 4
    Fix broken behavior before polish
  • 5
    Checkpoint before any large change
16
20 min / 20:10-20:30
Block 05

Publish & demo prep

Test it like a stranger would, then get ready to show it.
17
Publish & demo prep
20:10-20:30

Checklist

  • Open the published URL in a private / incognito window
  • Repeat the full value loop there, not just in the editor preview
  • Check the main screen at mobile width
  • Confirm nothing secret or private is exposed
  • Prepare a 45-second demo: problem, user action, useful result, next experiment
18
30 min / 20:30-21:00
Block 06

Lightning demos & Q&A

3-4 teams demo on stage. Everyone else: feedback, networking, and one actionable next step for your own idea.
19
Closing
20:30-21:00

Thank you for building with us tonight

Prompt pack
aisandboxlab.com/imaguru-sprint
Feedback form
menti.com/aljywq453jff
Your next step

Run your validation kit with a real person this week.

QR code to the prompt pack
Prompt pack
QR code to the feedback form
Feedback form
Vitalijs Visnevskis
axwise.de linkedin.com/in/vitalijs-visnevskis vitalijs@axwise.de
QR code to Vitalijs Visnevskis on LinkedIn
LinkedIn
20
Speaker notes
Arrows / Space: navigate  ·  N: notes  ·  P: print mode  ·  F: fullscreen