AI consulting in Ukraine: when you need it, how to choose, what to launch first
Cases from 2026 practice: SmileClinic (demo pipeline), Zoho migration, AI stylist. Market budgets, pitfalls, where to start. No hype, no filler.
Andrew Maryasov — AI consultant, founder of Auspex (CRM automation) and Grow2.ai (AI agents). 25+ years in business automation: accounting → CRM → AI. Builds strategy personally with owners and teams; implementation is run by his brands’ teams.
What “AI consulting” means in the Ukrainian context of 2026
The Ukrainian AI-consulting market has 3 distinctive traits:
- CRM fragmentation. Bitrix24, KeyCRM, Zoho, 1С/BAS, sometimes in-house systems. Any AI project starts with a CRM audit.
- Budget sensitivity. The average SMB budget is 3–5 times smaller than a Western competitor’s. This breeds discipline: ROI has to show up within 90 days, not within a year.
- Bilingualism. Most clients are Ukrainian-/Russian-speaking, and client-supplied content is mixed. RAG and FAQ agents need localized embeddings.
Who needs an AI consultant
A business benefits from AI consulting if:
- The team is >10 people and there are routine processes (support, lead generation, screening).
- There’s a CRM (any at all) — otherwise the first step isn’t AI, it’s a proper CRM.
- They’re ready to assign an internal product owner for 8–12 hours/week.
- They can name 3 metrics by which success will be judged.
If even one of these is missing — it’s still too early (and a good consultant will tell you so honestly at discovery).
Top 3 typical projects in 2026
The budgets below are market benchmarks for quality custom implementation in Ukraine, so you can calibrate vendors’ proposals. This isn’t my price list: implementation is run by the Auspex and Grow2.ai teams, and the estimate comes after discovery for a specific task.
1. AI agent for support or first contact
For whom: clinics, e-commerce, SaaS with >100 requests/day.
What it gives: −40 to −70% of the team’s time on FAQ, night+weekend 24/7 coverage, escalation of complex cases to humans.
Market budget: $4,000–$12,000.
Typical payback: 3–5 months.
See it from the inside: SmileClinic — a demo AI pipeline for a dental network. An honestly labeled demo: a full proposal package in 6 minutes, with calculations for a typical clinic.
2. CRM migration + AI fulfillment automation
For whom: businesses of 20–100 people where the current CRM doesn’t scale (Bitrix24 with no structure, Excel instead of a CRM, or a custom system with technical debt).
What it gives: a transparent funnel, lead scoring, an AI assistant for the manager, automatic follow-ups.
Market budget: $8,000–$25,000.
Typical payback: 4–8 months.
A real case: Zoho migration. Moving a financial calculator off Zoho Creator onto a modern in-house stack — in a few days instead of months of tender estimates.
3. AI stylist / cross-sell for e-commerce
For whom: clothing/decor/cosmetics stores with >500 SKUs.
What it gives: upsells based on a stylist’s logic (collections, seasonality, timing), personalized recommendations, an interactive UX instead of static cards.
Market budget: $6,000–$15,000.
A real case: AI Cross-sell Stylist. A stylist for a network of 30+ stores, built by the Grow2.ai team.
What the work process looks like
Step 0: Discovery (free, 30 min)
I listen about your business, ask the uncomfortable questions, and tell you straight: is there anything here for AI at all. Quite often the honest advice is to set up a proper CRM first.
Step 1: Strategy (with me personally)
A process audit, an applicability map (“where AI gives leverage, where it’s just noise”), a prioritized roadmap for 6–12 months, an implementation brief. Formats and prices are on the advisory page.
Step 2: Implementation (Auspex / Grow2.ai teams)
Architecture → MVP → CRM integrations → regression tests → launch on part of the traffic → scaling. The brief from the strategy is handed to the team directly, so you won’t have to explain everything again. A typical MVP at the implementer’s takes 2–3 weeks.
Step 3: Maintenance
The implementation teams handle maintenance and the evolution of the solution. I stay in touch at the strategy level — that’s what the advisory retainer is for.
The stack and tools I use
| Category | Tool | Why |
|---|---|---|
| LLM | Claude Sonnet 4.6 | The best accuracy in Ukrainian among frontier models |
| Vector DB | pgvector (Postgres) | No vendor lock-in, cheap, enough for most SMBs |
| Orchestration | Claude Code + n8n | Claude Code for agent logic, n8n for no-code chains |
| Frontend | Astro / Next.js | Astro for content, Next for interactive |
| CRM integrations | Bitrix24, Uspacy, amoCRM, Salesforce | Via API — Auspex’s profile |
It’s a pragmatic stack. If you already have a different one — I adapt, I don’t force a switch for the sake of switching.
Conditions you should require from any vendor
- NDA — signed before discovery, not after.
- Data — a clear answer on where it’s stored and whether it’s used to train models.
- Code ownership — everything done on the project belongs to you: the code, the prompts, the architecture, the documentation.
- An honest “no” — a vendor who never says no is selling you their hours, not a solution.
Who am I to write about this
28 AI projects across 14 industries and 7 countries. 8 of my own AI systems in daily production. Founder of Auspex (1,200+ CRM projects) and Grow2.ai. I advise from the position of someone who builds this with their own hands — and I honestly say no when AI won’t deliver ROI.
Free 30-minute discovery call →
FAQ
How is AI consulting different from AI development?
Consulting comes before development: discovery, architecture, ROI estimate. Without it, development is a lottery. A consultant may advise not to launch AI at all if the ROI doesn’t add up.
Do you do the implementation yourself?
I build the strategy and the architecture of the solution. The implementation itself is run by my brands’ teams: Auspex (CRM and processes) and Grow2.ai (AI agents) — with a ready brief from the strategy. If you already have a technical team, I can come in only as an architect / strategy consultant.
Do you work with 1С, BAS, Bitrix24?
CRM is Auspex’s home turf: Bitrix24 (cloud and on-premise), Uspacy, amoCRM, Monday.com, Salesforce. 1С/BAS — selectively, through integration wrappers. Exotic systems — an honest assessment at discovery.
What if we’re already working with another agency?
I can work as a second opinion at the architectural level, or take on one specific strategy module. I don’t claim to be the “sole vendor.”
How long until the first result in production?
The market norm for an MVP agent is 2–3 weeks at the implementation team. The first real metrics come a few weeks after launch on part of the traffic.
A letter on AI — 2–3 emails a month
Personal essays on how AI changes the way we think and work: what I tried myself, what worked and what didn't. At the end, links to the best of what I read.
No spam. Unsubscribe in one click.