01 What we do
A clear next step. At every stage.
Five capabilities, one team. Start where you are — a question, a pilot that stalled, or a system that needs to scale — and we’ll take the next step with you.
01Decide — AI strategy & discovery
Start with a better question.
Where can AI make a useful difference — and where is it a distraction? We map real workflows, measure today’s baseline, and help you choose what’s worth building first.
You might be asking
- “Which of our processes are ready for AI?”
- “Why did our last pilot stall?”
- “Build, buy, or wait?”
What you get
- Opportunity map scored for value and feasibility
- Baseline measurements for the top workflows
- Data, access and risk readiness review
- A scoped roadmap and a clear go/no-go decision
Typical formatExecutive Build Day or Discovery Sprint · 1 day to 2 weeks
02Build — Custom AI & software
Make the idea work in practice.
From a knowledge assistant to a customer-facing product, we design and engineer software around the job it needs to do — with the evaluation, interfaces and integrations that make it usable on a Monday morning.
You might be asking
- “Can an agent handle our first-line requests?”
- “Can this spreadsheet process become a product?”
What you get
- Bounded agents and workflow applications
- AI features inside your existing products
- Internal tools, customer apps and APIs
- Human review screens and audit trails
Typical formatPilot to Production · 6–10 weeks per workflow
03Connect — Data & AI foundations
Give intelligence the right context.
Good answers need the right data, the right permissions and the right model. We connect approved sources, choose models on evidence, and build the evaluation that keeps quality visible.
You might be asking
- “Is our data good enough?”
- “Which model should we use?”
- “Can this run on our own servers?”
What you get
- Permission-aware retrieval and data pipelines
- Model selection against your own examples
- Evaluation sets and acceptance thresholds
- Deployment in your cloud or on-premises
Typical formatPart of a pilot, or a 4–8 week foundation project
04Enable — Team enablement & transformation
The future works when people do.
Technology changes nothing until routines change. We work with leaders and teams to make new tools part of everyday practice — learning around real work, with the people who own it.
You might be asking
- “We bought licenses — why doesn’t anyone use them?”
- “How do we train 400 people without losing a month?”
What you get
- Executive working sessions
- Role-specific training for teams
- An AI champions program
- New operating routines and clear ownership
Typical formatExecutive Build Day, 4–6 week cohorts, or a 90-day champions program
05Improve — Reliability & continuous improvement
Keep learning. Keep it useful.
Launch is a starting point. We help you watch how the system performs, respond when models or data change, and choose the next improvement with evidence.
You might be asking
- “Who fixes it when quality drops?”
- “Why did our AI bill double?”
What you get
- Monitoring, cost controls and regression checks
- Incident runbooks and handover documentation
- Quarterly improvement reviews
- A support agreement scoped to your needs
Typical formatEmbedded AI Partnership · monthly
06 Ways in
Choose the right way in.
Four formats, drawn to the same time scale — from one day to a standing partnership.
- 01
Executive Build Day
1 day · For leadership teams
One day with your leadership team. We build a working prototype on one of your real workflows, together, and leave you with three prioritized opportunities.
- 02
Discovery Sprint
≈2 weeks, confirmed after scoping · For teams with a candidate use case
A focused assessment of one or two workflows: baseline, data and access review, risks, and a pilot plan with acceptance criteria.
- 03
Pilot to Production
≈6–10 weeks · For a validated opportunity
A paid, bounded pilot on approved data. When it meets the agreed evaluation and operational criteria, a separate production scope follows.
- 04
Embedded AI Partnership
Monthly · For organizations scaling beyond one workflow
A named team working alongside your internal owners, with monthly capacity, support scope and service levels agreed together.
07 Plainly
What we won’t sell you.
- A platform you have to adopt.
- Strategy that ends in slides.
- A pilot without acceptance criteria.
- Lock-in to one model vendor.
- Automation that removes people from decisions they should own.
08 Questions
Questions we hear first.
Do you work with our existing vendors?
Yes. We integrate with the systems you already run and work alongside your IT and data teams and your current vendors. We add only what a workflow needs, such as retrieval, orchestration, evaluation and monitoring, and document it so your team and vendors can support it. We don’t ask you to adopt a new platform.
Which AI models do you use?
The one that performs best on your examples, at a cost you can sustain. We work with Anthropic, OpenAI, Google and open-weight models, in the cloud or on your own infrastructure. Candidates are compared on the same evaluation set for quality, cost and speed, so the choice can be revisited when prices or models change.
Who owns what we build?
You do, by default. Ownership terms are agreed up front, and unless we agree otherwise you own the code, prompts, evaluation sets and documentation created for you. Everything is handed over with a runbook and training, so your team can run and improve the system without depending on us.
How do you price?
Fixed scope and fixed price for Build Days, Discovery Sprints and pilots. Monthly for partnerships. After a first conversation you receive a written recommendation that states the first step, its scope and what it costs, so the price is agreed before any work starts. A production scope after a pilot is priced separately, once the pilot has met its criteria.
Do you work in Turkish and English?
Yes. Our team works in both, and we build systems that handle both. Workshops, training, documentation and evaluation sets can be in Turkish, English or a mix, matching the language your teams and customers actually use every day, from internal policies to customer-facing replies.
How long does an AI pilot take?
About six to ten weeks for one workflow. A Pilot to Production engagement runs on approved data with acceptance criteria agreed up front: one to two weeks to align and validate, several weeks of weekly builds, a pilot on live cases and a handover. When the pilot meets its criteria, a separate production scope follows.
Where should we start if we are not sure?
Start with one conversation, then usually an Executive Build Day or a Discovery Sprint. The Build Day suits a leadership team that wants to see AI work on a real workflow in one day; the Sprint suits a team with a candidate workflow that needs a baseline and a pilot plan. We recommend one in writing after the first call.