archive · case_studies · receipts attached

Real systems. Real numbers.
Receipts attached.

Q: What is kratt?

kratt is a two-person AI consultancy that finds where a business loses revenue, then builds, hosts and runs the systems that close the gap. The case studies below list the systems that actually run, with the live URL attached wherever the work is a public site.

Q: How does kratt measure case study results?

Client case studies report the metric the client asked us to move, call response time, consultations booked, hours returned, revenue lifted, ad spend saved, measured against the baseline we record before the build starts. The systems we run on ourselves report numbers counted from our own logs, and each page lists what it does not claim. No vanity metrics, no cherry-picked windows.

Q: Will kratt share my case study publicly?

Only with written approval. Public case studies are reviewed line-by-line with the client and never go live without sign-off. Many kratt clients prefer to keep wins private, that is fine, we have a parallel library of confidential case studies shared only on strategy calls under NDA.

Q: Can I talk to a kratt reference client?

Yes, after the second strategy call. We introduce qualified prospects to a current or recent client in a similar vertical. References take a 20-minute call directly with you, no kratt operator on the line, no script, no filter.

Every kratt case study below is in production, with the live URL attached wherever the work is a public site. Named proof includes ProfiElekter, diil.ai and kratt's own internal systems, linked where they can be inspected publicly. Numbers we cannot re-measure are stated as claims we do not make, rather than dressed up as metrics.

See the live portfolio wall Every site we have shipped, live, on one page.

[ the_builds

Nine systems. Full receipts.

Case studies are reserved for AI, infra, search and operating systems. Plain website work stays in the portfolio wall; every card here has a concise description and a dedicated full-story page.

Local SEO System Electrical contractor · Estonia 01

ProfiElekter OÜ

Seventeen years of real projects (Swissotel, Hilton, a 50,000 m2 industrial LED refit) sat invisible on a 13-year-old Weebly site. We migrated it like-for-like first, losing zero ranking pages, then grew it into a 125-page multilingual search system. The useful proof is the technical SEO layer, structured documentation, and audit trail.

125

pages, 3 languages

96/100

technical SEO (A)

View live · profielekter.ee Full story →

Proptech Platform Proptech SaaS · diil.ai 02

diil.ai

A multi-tenant proptech platform where every property developer runs listings, leads and investor reporting on their own subdomain. Traditional agencies quoted $80-150K and 4-6 months. We built it as six modules on our own stack. It now runs $100M+ in transactions across 50+ agencies.

platform modules

$100M+

deals managed on it

View live · diil.ai Full story →

VibeCode Our own pipeline 03

The kratt lead engine

Manual prospecting capped our own pipeline at 20-30 conversations a month. So we built an engine that finds businesses, scores their websites with AI, and generates a personalized landing page per prospect. It feeds our own outreach for about $0.50 a lead.

200+

qualified leads / mo

<$0.50

cost per lead

kratt-built · our own outreach Full story →

AI Implementation AI SaaS · diil.ai 04

diil.ai SEO engine

A Python engine pulls keyword clusters, writes long-form posts in diil.ai's brand voice, and publishes to a Supabase-backed blog on a cron. It ships 100+ posts a month with zero human hours. The output is public, read it live.

100+

posts / month

human hours

See it live · diil.ai/blog Full story →

AI Implementation Creative production · kratt-built 05

The content machines

A four-loop pipeline that turns a codified brand DNA file into finished ads, carousels and campaign frames. It fingerprints every combination it fires and refuses to repeat one inside 30 days. 66 machine-rendered creatives across 6 brands so far.

creatives, all shown

720

composer combinations

kratt-built · the engine behind Ad Factory Full story →

CMS + Deploy Loop Leather craft · Estonia 06

Nahkur

A one-man leather workshop got a static site with an editor-to-production system underneath it. Catalogue data, Valencia product data, portfolio cards and hourly-rate copy became editable without turning every change into a developer ticket. The receipt is the CMS render loop, bilingual data model and deploy path.

CMS render passes

languages wired from data

CMS loop · public proof Full story →

Infra + Automation Kratt operations · internal 07

The automation spine

Kratt runs a shared n8n automation instance instead of scattering one-off scripts across laptops. It sits behind a public webhook front door and a private admin path, with Postgres state, execution pruning and runbook limits. The public story is architecture and operating discipline, with secrets and endpoint details deliberately withheld.

workflows present

active workflows

14d

execution retention

internal infra · source card K-02 Full story →

Data + Lead Ops Estonian open data · internal 08

Registry radar

A daily Ariregister diff watches Estonian company data for new registrations, new VAT numbers and liquidation signals. A second list-hygiene job reconciles the live call list against that snapshot so sales knows who to call and who not to touch. It runs read-only until a human decides the signal quality is worth writing into CRM.

376k+

entities diffed

unit tests

219/292

first hygiene matches

internal data system · source cards B-11/B-12 Full story →

Infra + Recovery Kratt operations · internal 09

Backup and restore test

The backup loop dumps the operational stores that would hurt if the box died, then verifies row counts instead of trusting file existence. A separate weekly restore test loads those dumps into a throwaway Postgres and destroys it after proving the data loads. It is continuity work with failure alerts, retention rules and known gaps stated plainly.

stores covered by dump

14+8

daily + weekly retention

production writes during restore test

internal infra · source cards K-16/K-17 Full story →

[ next_build

Want to be the next case study?

30-minute call. Screen on. We map your stack live, find the 3 highest-ROI loops, and tell you what they'd cost to ship. The audit is free.

[ book_audit ]

Case Studies · FAQ

How we report results, and how to vet us.

The metric we use, the consent rules around publishing, and how to talk to a reference client. Three answers most procurement teams ask before they buy.

What is kratt?+

kratt is a two-person AI consultancy that finds where a business loses revenue, then builds, hosts and runs the systems that close the gap. The case studies below list the systems that actually run, with the live URL attached wherever the work is a public site.

How does kratt measure case study results?+

Client case studies report the metric the client asked us to move, call response time, consultations booked, hours returned, revenue lifted, ad spend saved, measured against the baseline we record before the build starts. The systems we run on ourselves report numbers counted from our own logs, and each page lists what it does not claim. No vanity metrics, no cherry-picked windows.

Will kratt share my case study publicly?+

Only with written approval. Public case studies are reviewed line-by-line with the client and never go live without sign-off. Many kratt clients prefer to keep wins private, that is fine, we have a parallel library of confidential case studies shared only on strategy calls under NDA.

Can I talk to a kratt reference client?+

Yes, after the second strategy call. We introduce qualified prospects to a current or recent client in a similar vertical. References take a 20-minute call directly with you, no kratt operator on the line, no script, no filter.