AT.
ENRUUABY

Got a process that eats hours every week?I'll build the AI tool that takes it over.

2-6 weeks, fixed price, money back if it doesn't work.

What I've shipped

MarketingFinanceAI safetyB2C SaaS
01

Agentic-Email-Outreach-OS

Want one for your business?

Cold but personalized B2B outreach that gets answered.

The problem

A boutique AI video-production studio in Warsaw. The founder's personal cold email worked — but he runs the studio, and writing each one by hand just isn't realistic. Mass template blasts get spotted instantly. Without personalization, you get about 1‑3% replies (Belkins data, Instantly benchmark).

What I built

  • Researches every lead — one specific recent project as the hook, the email in the founder's voice.
  • Trained on the founder's voice — 28 voice rules, 17 industry maps, a check before send. Reads as hand-written, except there are hundreds of them.

Result

7% reply rate across 500+ emails (10% in the US, 8% in the UK).

Tech
Apollo, Instantly, Claude API
Status
Live for a Warsaw AI video-production studio.
Marketing opsCold outreachUSUK
02

svor.ai

View live →

AI sorts every transaction by project and counterparty; P&L, Balance and Cashflow update in one window.

The problem

Growing SMBs run accounting across five disconnected tools — Pyrus, Xero, Excel, Telegram, a custom CRM. Reconciliation eats days.

What I built

  • Multi-bank parser pipeline — 5 Ukrainian banks, cp1251 encoding.
  • AI auto-matches every transaction to a project and counterparty; live P&L, Balance and Cashflow.

Result

85% auto-match accuracy on real financial data.

Tech
FastAPI, Python, Claude API, React
Status
Live, in pilot.
Finance opsReconciliationUA
03

SafeRun, AI agent safety stack

View live →

Runtime safety for AI coding agents in production.

The problem

AI coding agents (Cursor, Claude Code, Copilot) run git push --force or git reset --hard on their own. One wrong call rewrites history — and you can lose the project.

What I built

  • 3-layer interceptor — shell wrapper, Git reference-transaction hook, GitHub webhooks.
  • Caught before execution — dangerous commands go to Slack with a risk score and one-click Approve / Deny. AES-256-GCM, local-first.

Result

Shipped and installable — npm i -g @saferun/cli (v0.6.63). Intercepts in real time. Submitted to the Anthropic Plugin Directory.

Tech
Bash, jq, Node.js, TypeScript
Status
Open source. npm live.
AI safetyDevSecOps
04

AI Translator for Elementor and Polylang

View live →

A WordPress plugin that translates your site without breaking the Elementor layout.

The problem

Translating a WordPress site usually collapses the Elementor layout — widgets shift, sections break, and someone rebuilds the page by hand.

What I built

  • Translates while keeping every widget and section pixel-perfect.
  • AI translation layer on the WordPress + Elementor + Polylang stack.

Result

470+ downloads and a paying tier — shipped to WordPress.org.

Tech
PHP, OpenAI API, WordPress plugin API
Status
Shipped to WordPress.org, reached a paying tier.
B2C SaaSWordPress

How we work together

Four steps from first call to a working tool. Intro call → brief → audit → build (if it makes sense).

StepDurationPriceWhat you get
1. Intro call30 minFreeFree call. We check fit and get a rough sense of the problem.
2. Brief + NDA1-2 days—We talk again — I dig into the process and what the audit will cover. We agree the audit's scope and price (3,500-7,000 PLN), sign NDA. 50% deposit.
3. Audit7-14 working days3,500-7,000 PLNI work through your process in sessions with you and whoever runs it. You get: process map, bottlenecks, working prototype of 15-20% of the solution, written recommendation — build, partial, or don't build.
4. Build (if recommended)2-6 weeks8,000-25,000 PLNWorking tool deployed. Methodology docs. Runtime spec library. Git repo. 30-day warranty.
1. Intro call
Duration
30 min
Price
Free
What you get
Free call. We check fit and get a rough sense of the problem.
2. Brief + NDA
Duration
1-2 days
Price
—
What you get
We talk again — I dig into the process and what the audit will cover. We agree the audit's scope and price (3,500-7,000 PLN), sign NDA. 50% deposit.
3. Audit
Duration
7-14 working days
Price
3,500-7,000 PLN
What you get
I work through your process in sessions with you and whoever runs it. You get: process map, bottlenecks, working prototype of 15-20% of the solution, written recommendation — build, partial, or don't build.
4. Build (if recommended)
Duration
2-6 weeks
Price
8,000-25,000 PLN
What you get
Working tool deployed. Methodology docs. Runtime spec library. Git repo. 30-day warranty.

Why this audit is worth what it costs

Three things you're paying for. Not consulting hours.

01

I find the real problem

Nine years in product marketing — fintech and adtech, dozens of customer interviews. I've learned to tell a real pain from something that only looks like a problem. That's where the audit starts: working out what's actually worth fixing, and what isn't.

02

I design it and build it

But I don't stop at the diagnosis — I build the fix myself. Built a marketing department from 0 to 20 people, and I write the code: my own WordPress plugin and a DevSecOps tool, end to end. So on the audit I see right away what's actually buildable in 2-6 weeks and what isn't. No broken telephone between the business and the code.

03

Done it, not just advised

And none of this is theory. I took a fintech platform to profitability in six months by pivoting it from B2C to B2B. I relaunched a stalled adtech product after interviews showed what the market was missing. I come to your problem with that, not a blank page.

I reserve the right to recommend no-build

Roughly 1 in 5 audits ends with me recommending NOT to build the tool. You get the audit deliverable either way.

Sometimes a workflow change solves the problem. Sometimes an existing SaaS does it cheaper. Sometimes the ROI doesn't justify the build cost. In all those cases I'll say so, with reasoning, in writing. Better to find out before you spend 20,000 PLN on a tool nobody will use.

What's included in each tier

Audit and Build both deliver portable artifacts you own. Here's what comes with each one and what doesn't.

AuditBuild
Duration7-14 working days2-6 weeks
Price3,500-7,000 PLN8,000-25,000 PLN
DeliverableRecommendation + working prototypeWorking tool, deployed
RiskLow. Recommendation may be 'don't build'.Money back if it doesn't work as agreed.

Audit deliverables

PLN 3,500-7,000
  • Process map of current workflow
  • Bottleneck analysis. Hours lost per week, named bottlenecks.
  • Working prototype of one slice (15-20% of full solution)
  • 3 solution variants with trade-offs
  • Written recommendation: build, partial, or don't build
  • If build is recommended: fixed-price proposal for next stage

Build deliverables

PLN 8,000-25,000
  • Working tool, deployed where agreed
  • Methodology documentation. How the system works and why.
  • Runtime spec library (prompts, agents, rules, templates)
  • Voice / brand training (if applicable)
  • Examples library (anonymised real cases)
  • Git repository, private, ownership to you
  • README, setup guide, architecture diagram
  • 30-day bug-fix warranty
  • 1-2 hour handover session

Not included in Build

Discussed separately
  • Production hosting on your infrastructure
  • Long-term maintenance after the 30-day warranty
  • Integrations not scoped in Audit
  • Training for more than one person (default: 1 person handover)
  • White-label or resale rights

About me

Casual photo of Artem Taranenko

Senior Product Marketing Manager with 9 years in B2B SaaS, mostly fintech and adtech. Built and led a 20-person product / marketing org at Uniqit Group. Mini MBA with Mark Ritson, 2024.

Now solo. I run discovery, scope the problem, and build the tool myself. No handoffs. Most AI implementers know the tech but not the business. Most consultants know business but can't build. I do both. I keep projects small enough to ship in 2-6 weeks.

LinkedInGitHub

What usually gives pause

Honest answers to the usual doubts.

You're one person — what if you disappear?

You're not tied to me. The code and access are yours from day one, everything's documented (README, methodology, architecture), and it's built on a standard stack any developer can pick up. Projects stay within 2-6 weeks, plus a 30-day warranty and optional support after.

What counts as 'it worked'?

On the audit we write down 1-3 measurable success metrics — that's the acceptance test, not whether I happen to like it. If they're not met after the build, I refund the build.

I pay for the audit — what if it says don't build?

A written case for why not — process map, bottlenecks, and numbers you can verify yourself. The 3,500-7,000 PLN audit just saved you 8,000-25,000 on a tool nobody would use. And I'll tell you what to do instead.

What if the AI gets it wrong on important data?

The tool is an assistant, not the source of truth — a human has the final say. I build in discrepancy checks and a clear path for when the model isn't sure. No blind trust.

Is this safe? Where does my data go?

Raw sensitive data never goes to the LLM — I pseudonymise identifiers, strip account and document numbers, and bucket amounts (that's how the privacy layer in svor.ai works). Models aren't trained on your data. NDA up front; hosting in an EU region where available, full EU residency on request (Claude via AWS Bedrock / Google Vertex). DPA and a data-flow map at the start.

Common questions

What if you tell me not to build?
You still get the audit deliverable. Process map, bottleneck analysis, written recommendation, and the working prototype. About 1 in 5 audits ends this way. The audit is paid regardless of build outcome.
Do you only work with AI projects?
I build AI-powered tools, but the audit often surfaces non-AI fixes too. If your process can be solved with a spreadsheet template or a workflow change, I'll say so. I'm not married to 'must use AI'.
What languages do you work in?
English, Russian, Ukrainian for delivery. Basic Polish for written communication. Contracts in English or Polish.
How do you handle data privacy?
Raw sensitive data never goes to the LLM — I pseudonymise identifiers, strip account and document numbers, and bucket amounts (that's how the privacy layer in svor.ai works). Models aren't trained on your data. NDA before any project; hosting in an EU region where available, full EU residency on request (Claude via AWS Bedrock / Google Vertex). DPA and a data-flow map at the start. GDPR / RODO.
What's the payment structure?
Audit: 50% upfront, 50% on delivery. Build: 50% upfront, 50% on handover. Money back if the Build doesn't work as agreed (defined in contract).
Do you take retainers?
Optional. PLN 2,000-4,000 / month for ongoing changes and monitoring. Not required.
What's not included by default?
Long-term hosting, integrations not scoped in Audit, training for more than one person, white-label or resale rights. All discussed separately if needed.

Got a process that's eating hours every week?