Skip to main content
veles

// Process

How we get from your idea to a number we both trust

Veles does not publish tiered pricing. Every custom AI build varies too widely for that. This page walks you through the four stages of a typical engagement and the scoping methodology we use to size each project honestly before any code is written.

The short version

Four stages: Discovery, Scope and Estimate, Build, Launch and Operate.

Typical end-to-end is 2 to 8 weeks. Smaller integrations land in 2-3 weeks. Larger multi-channel systems run 6-8 weeks.

After a 30-minute discovery call you get a written summary, a milestone-based estimate, and a fixed price for the agreed scope. No vague "starting from" numbers, no surprise invoices, no padding.

Engagement models: fixed-scope project (typical) or operate-and-iterate monthly retainer (for clients who want us to keep tuning the system post-launch).

1. Discovery

A 30-minute call. Free, no commitment. You walk us through the workflow, the volume, and what you actually need the system to do. We listen, ask the questions that surface the hidden complexity (every project has some), and write up what we heard.

What you get from Discovery:

  • A written summary of the problem in plain language, so we both agree on what we are solving before anyone scopes it
  • A short list of the hard parts. The things that turn a "two-week build" into a "six-week build" if missed early
  • A direct honest answer: yes this is a good fit for custom AI, or no, here is a no-code path that will save you money

If we are not the right fit we will tell you so on the call. We have turned down work that would have been profitable for us but wrong for the client. The 30 minutes are worth it either way.

2. Scope and Estimate

After Discovery we go away and map the project into milestones. This usually takes 2 to 5 working days depending on complexity. You get back a written proposal with:

  • Milestones, each with a clear deliverable, dependencies, and an estimated effort range in working days
  • A fixed price for the agreed scope. Not "starting from", not "+/- 20%", a real number we both sign off on
  • What is included: every assumption made, every integration touched, every edge case considered
  • What is NOT included: every assumption that did not get made, every integration deferred, every nice-to-have parked for v2. We surface these explicitly so you are never surprised
  • A delivery timeline with weekly milestones and a target launch date

If something in the proposal does not sit right, we iterate. The goal is one round of revision, then a contract.

3. Build

A small senior team works in weekly sprints. You get a live demo at the end of each week, not a polished slide deck. You can ask for changes mid-build without paperwork (small changes inside the agreed scope are absorbed; out-of-scope changes get a brief written addendum).

Build discipline:

  • We write real code. No no-code stack repackaging, no Zapier with a markup
  • Every milestone goes through internal code review before you see it
  • Production data is touched only after we both sign off on the data processing approach
  • AI behavior is reviewed before customers see it. We do not push agents into production and hope

You see progress every week. There are no two-month dark periods where the team disappears and reappears with a "demo coming soon" message.

4. Launch and Operate

We deploy to production. The first two weeks are close monitoring: we tune thresholds, fix the edge cases that only show up under real load, and patch anything the dry-runs missed. This work is included in the fixed-scope build.

After the first two weeks, two options:

  • Hand-off: the code is yours, documented, with a runbook. We are available for paid upgrades if something needs changing later. You call us when you need us
  • Operate-and-iterate retainer: a predictable monthly rate where we keep tuning the system, monitor for drift, update integrations when third-party APIs change, and ship small improvements every month. Best fit for AI systems that face changing inputs (customer conversations evolve, market conditions shift, etc.)

Both options are good. Most clients start with hand-off and move to retainer after 1-2 months if the system needs ongoing care.

How we estimate

We scope every project, we do not price-list it. Here is what goes into a Veles estimate:

  • Engineering hours per milestone, costed at our blended rate. The rate is the same whether the work is API integration, agent design, or dashboard build. Senior team, no junior dev markup
  • Integration cost: each external system (CRM, scheduling, messaging API, billing) carries a baseline integration effort. We have done most of them before, so the estimate reflects real prior data
  • AI behavior tuning: agents need iteration. We budget explicit cycles for behavior tuning in the build phase. This is where over-optimistic vendors burn through their margins; we name it upfront
  • Buffer for the unknown: every project has surprises. We add a small explicit buffer for them. If the buffer goes unused, the price stays the same

The estimate is fixed for the agreed scope. If you change scope mid-build, we write a short addendum with the additional milestone and price. No silent surprise invoices.

Why no tiered pricing

Pricing pages with "Starter / Pro / Enterprise" tiers work for productized services where every customer gets the same thing. Custom AI is the opposite. A focused single-channel AI agent costs meaningfully less than a multi-channel automation with custom CRM integration. Putting both on the same "Pro tier" would be lying to you.

The honest answer is: tell us what you need, give us 30 minutes, and we will write you a real number based on what we actually heard. The tier-card sites are guessing. We refuse to.

// FAQ

Frequently asked

What does a Veles project typically cost?
It depends on scope, and that is not a cop-out. A focused single-channel AI agent (one input source, light CRM integration, no custom dashboard) is on one end of our range. A multi-channel automation with custom data pipelines, on-premise inference, and an operate-and-iterate retainer is on the other end. After a 30-minute discovery call you get a real, fixed-price estimate. We do not give "starting from" numbers because they are misleading.
How does the 30-minute discovery call work?
Free, no commitment. You book a time on Calendly, we get on Google Meet, you tell us what you are trying to do and what you have tried so far, and we ask the questions that surface the hidden complexity. By the end of the call we both know whether Veles is the right partner for this project. If we are, we go away and write you a scope proposal within 2-5 working days. If we are not, we will tell you on the call.
What if we change scope mid-build?
Small changes inside the agreed scope are absorbed at no extra cost. That is what the buffer is for. Larger out-of-scope additions get a short written addendum: new milestone, new line-item price, your sign-off, and then we build it. There are no surprise invoices and no charge for changes you did not approve.
Do you sign a data processing agreement?
Yes. We are EU-based (Belgrade, Serbia) and GDPR-compliant by default. Before any production data moves we sign a Data Processing Addendum (DPA) covering scope, retention, sub-processors, and breach notification. For sensitive workloads we can run inference on-premise or via privacy-preserving providers, so your data never trains anyone else's model.
Who actually does the work?
A small senior team. The same engineers who scope the project also build it and operate it post-launch. We do not "sell senior, deliver junior." When you have a question six months after launch, the person who answers is the person who shipped the code.
What if the project is too big for a fixed price?
We split it. The first milestone is always self-contained and shippable. If the project is too large to scope in one shot, we run a paid Discovery Sprint (1-2 weeks, fixed price) at the start that produces an architecture document and a phased estimate. From there each phase is its own fixed-scope build. This keeps risk low on both sides.
Do you take equity or revenue share instead of cash?
Rarely. For most engagements we work on a fixed-scope cash basis. For specific projects that align with our long-term focus we are open to discussing partial equity or revenue share alongside a reduced cash component, but the conversation starts with cash and an honest assessment of fit.
Can you work with our internal team?
Often. We have shipped projects where Veles owns the AI layer and the client team owns the rest of the stack, with us pair-programming on the integration points. We have also handed off projects where the client team takes over operations post-launch and we stay available for monthly check-ins. The engagement model is part of Scope and Estimate.

Ready to scope something?

Book a 30-minute discovery call. You walk through the problem, we ask the questions that matter, and we leave the call agreeing on whether Veles is the right partner for it.

book a discovery callOr email us first

We use cookies to analyze site traffic and improve your experience. Privacy Policy