Straight answers

Fixed-fee for scoped work (audits, builds, sprints), monthly for fractional retainers, and a day/hour rate for ad-hoc advisory. Published prices are 'from €X' starting points; the exact number comes out of a short scoping call once the work is clear.

A fixed-fee Architecture / AI-Readiness Audit (from €8,000, ~2 weeks) or a PoC Sprint (from €9,000, 2–4 weeks). Both give you something concrete — a roadmap or a working proof-of-concept — without committing to a full build first.

Yes. I'm Israel-based and work with clients across the EU, UK, US and Israel, in English and Hebrew. For US-headquartered clients I bill in USD; otherwise EUR.

Answers are grounded in your documents with citations, critical facts are validated before they're shown, and an eval set scores answer quality and hallucination rate continuously. When a wrong answer appears, the retrieval/model/data split is diagnosable rather than a black box.

Yes, with care. Hebrew brings real constraints — RTL, morphology, and tokenizer coverage that inflates cost and degrades quality — that break naive pipelines. I choose embeddings and parsing for the language and validate retrieval and answer accuracy on your own Hebrew content before deploying.

Wherever you need it to. The vector database, embeddings and retrieval can run entirely inside your VPC or on-prem, with private endpoints and your own encryption keys — nothing shipped to a third-party API. Per-tenant access control keeps users from retrieving documents they shouldn't see.

No — not on Commercial/Enterprise terms, and not via the API or Amazon Bedrock. I'll show your security team the exact contractual and technical settings that guarantee it, which is the difference between the enterprise tier and the consumer plan they're worried about.

Yes — Claude Code via Amazon Bedrock with VPC endpoints keeps prompts and code on a private network path inside your AWS account boundary, with IAM-scoped access and no public internet egress. Initial deployment is roughly a few hours of admin work, not a multi-month project.

Humans keep the gates. I set up protected branches, CI/CD approval gates and a clear RACI — who owns repo risk tiers, review gates and release approvals — so AI-generated code still passes through the same controls as anything else.

Default to keeping data inside your boundary: VPC/on-prem deployments, no-training terms, private endpoints, and protected-branch workflows. The whole point of the Bedrock/VPC and self-hosted-embedding patterns is that sensitive code and documents never leave your control.

An audit is ~2 weeks, a PoC sprint 2–4 weeks, and a RAG or computer-vision build 4–8 weeks depending on scope. Every fixed quote is anchored to a locked scope and a short discovery step — scope creep is what blows fixed prices.

You get the system, the eval harness and the documentation to run it. I can stay on as a fractional architect for ongoing direction and reviews, or hand over cleanly — your choice, not a lock-in.