- Sheridan, Wyoming, US
- [email protected]
- Call Us Anytime+123 (4567) 890
- Email Us[email protected]
- Opening HourMon - Sat 8:00 - 6:30, Sunday - CLOSED
Get latest update for our trusted applications
FAQ
FAQ
Straight answers, including the awkward ones.
Everything below is what we would tell you in a first meeting. Where the honest answer is unflattering to us — the projects that failed, the fees we forfeited, the work we cannot do — it is on this page rather than omitted from it.
1 in 5
we declineAssessments concluding: do not proceed
2
failuresEngagements that missed their target
99%
retentionClients continuing after project one
Getting Started
How a relationship begins.
- 01. What does the first engagement look like?Almost always a two-week value assessment at a fixed fee, credited in full against anything that follows. It produces a written value model, a ranked opportunity list and a recommendation — which is sometimes not to proceed. You keep the document either way.
- 02. Can we skip the assessment and start building?You can, and we will usually argue against it. Two weeks establishing what the work is worth is what makes everything afterwards accountable. Where a client insists, we still write the value model, because we would rather know whether we succeeded.
- 03. How quickly can you start?Assessments typically begin within two to three weeks. Delivery start depends on the practice; AI and engineering capacity is usually available within four to six weeks of a signed engagement.
- 04. Do you work with small companies?Nearly a third of our clients have fewer than fifty people. What matters is whether a business number worth moving exists, not the size of the organisation around it. The scope and the fee scale down honestly.
Commercial
Money, risk and what happens if it fails.
- 01. How do you price?Fixed price for defined scope, monthly retainer for embedded teams, and outcome-linked with up to 30% of our fee at risk where the metric is clean enough to be fair. Delivery from Kathmandu lands around 30% below comparable onshore cost at the same seniority.
- 02. Will you guarantee the business outcome?We will contract a portion of our fee against it, which is as close to a guarantee as anyone honest can offer. We cannot guarantee outcomes that depend on your team's adoption and decisions, and we will not pretend otherwise in a proposal.
- 03. What happens when a project misses its target?Two engagements in our history have. In both, a portion of our fee was forfeited under the outcome agreement, and in both we wrote up what went wrong and shared it with the client's board. We will discuss both in a procurement conversation.
- 04. Why do you turn down work?Because about one in five assessments concludes that the proposed work will not pay for itself, and taking the money anyway would make every other claim on this website worthless. It has cost us revenue and produced our retention rate.
Fig. 01 — The monthly value review, where these answers get testedBaseline · actual · variance · next action
Ownership & Data
What you own, and where your data lives.
- 01. Who owns the code and the models?You do, from the first commit. Code sits in your repositories, infrastructure in your cloud accounts, models in your tenancy. There is no proprietary Golden layer and no licence you need to keep paying to use what we built.
- 02. Will our data be used to train models?Never anything that leaves your environment. We deploy with provider-side training and retention disabled, and this is contractual rather than a settings page. Your data is not used for other clients, other engagements or our own product development.
- 03. Can our data stay in our jurisdiction?Yes. Everything runs in your cloud accounts in your chosen region, and our Amsterdam presence supports EU-only processing where regulation requires it. Air-gapped deployment with open-weight models is available for the strictest cases.
- 04. What happens if we end the engagement?You keep everything and we run a documented handover. We hold no credentials, code or data as leverage. Our engagements are designed to be endable, which is one reason so few clients end them.
AI Specifics
The questions risk committees always ask.
- 01. How do you stop AI from hallucinating to our customers?Answers are constrained to retrieved, approved content with citations, confidence thresholds route uncertain cases to humans, and outputs are validated before they commit. Where a claim cannot be grounded, the system says so rather than inventing.
- 02. Can every automated decision be explained?Yes. Inputs, model version, policy version and reason codes are logged for every decision and can be replayed exactly as they ran. One client reconstructed an eighteen-month-old credit decision for a regulator in an afternoon.
- 03. Which AI models do you use?Whichever wins your evaluation set at acceptable cost and latency — OpenAI, Anthropic, Google, Llama, Mistral or a fine-tuned model. We build model-agnostic on purpose, because the frontier moves every few months.
- 04. Will AI replace our staff?In our deployments it has redistributed work far more often than removed people: agents absorb volume growth and repetitive processing, humans move to exceptions and relationships. We model the people impact honestly in the assessment rather than after signature.
- 05. Are you compliant with the EU AI Act?We classify each use case by risk tier and build the corresponding obligations into delivery — documentation, data governance, human oversight, logging and transparency. Compliance is a property of your system, and we build it to hold.
Working Together
Teams, geography and time zones.
- 01. Who will actually do the work?Named individuals listed in the proposal, including the senior engineer who scoped it. We do not maintain a graduate bench billed as senior capacity, and we do not substitute seniority downwards without telling you why.
- 02. How do time zones work?Engineering is centralised in Kathmandu with defined overlap hours written into each working agreement. European and Gulf clients have substantial natural overlap; Americas clients are served by a shifted team plus 24/7 support.
- 03. Can you work alongside our internal team?It is our preference. We pair with your engineers, adopt sound existing conventions and aim to leave the team more capable. Several clients now operate everything we built with no external help.
- 04. Do you take over work from another vendor?Regularly, following an honest assessment of what is salvageable — which occasionally concludes that continuing costs more than restarting. Either way you get the reasoning in writing.
Not Answered?
Ask us the harder version.
The questions worth asking a technology partner are usually the ones that are awkward to answer: what have you failed at, what do you earn from your recommendations, who owns the work if we fall out, and what happens if the number does not move.
All four have answers on this site, and we would rather you asked them directly. A partner who becomes uncomfortable at that point is telling you something useful.
Read Next
Read next
How we measure value
The value model, the four levers and the mechanisms that stop this being a slogan.
EXPLORE MOREEXPLORE MORERead next
How we price
The ROI arithmetic, what counts as a benefit and what we deliberately exclude.
EXPLORE MOREEXPLORE MORERead next
How we work
Analyse, build, grow — and what the engagement requires from your side.
EXPLORE MOREEXPLORE MORE1 in 5
Assessments say stopWritten recommendations not to proceed
2
Engagements missed targetFees forfeited in both
99%
Client retentionClients continuing after project one
Related
Where to go next.
Next Step
Bring the question this page did not answer.
A first conversation is with the practice lead and senior engineer who would run the work. Ask them anything, including what they have failed at.
