Where experience meets execution.
The people who were early to understand search algorithms in 1999 are among the first to understand answer engines in 2026. We help organizations decide where to deploy AI so that it actually works in daily operations, long after the board meeting where it sounded good.
The last two years changed the rules. Organizations that are not integrating AI are already losing operational ground this quarter, and the opposite is equally true: organizations that move into AI without a plan burn an annual budget on impressive pilots that never become deployments, and discover late that the real constraint was access to data and to core systems.
In a large organization that gap is wider still, because every deployment touches regulation, information security, and processes that cannot be paused. We come in at the stage right before the investment, the one that determines whether it pays back.
Banks, insurers, investment houses and fintech companies operating under supervision, that need to deploy AI without compromising reporting and control obligations.
HMOs, hospitals and medical institutions, where data sensitivity dictates both model selection and deployment architecture.
Ministries, authorities and statutory corporations, with specific transparency, accessibility and procurement requirements.
Organizations with legacy core systems, multiple business units, and operations across several languages.
If your organization is smaller than 20 people, full strategic consulting is probably not the right instrument, since a team workshop or a focused automation will return value faster, and we will say so on the first call.
A two-hour kickoff, interviews with three to five key people, process mapping, a review of existing systems and a check of which ones are genuinely reachable via API. The output is a ranked opportunity matrix with expected return for each, plus a decision document with a reasoned recommendation.
Selecting the first pilot from the matrix, specification with KPIs and a timeline, AI vendor selection, an information security framework, and a measurement plan. The output is a specification ready for development, with us or with your own team.
Guiding the pilot until it produces measurable value, a biweekly review call, real-time adjustments and gradual expansion. You end with a live system that works, is measured, and can be replicated to further processes.
Your rulebook: what an agent may do, who approves, what is logged, and exactly where a human stays in the loop.
The measurement instrument that precedes the investment, reviewing technology, people, processes and data, and producing a readiness score from 0 to 100.
A plan explaining where AI will deliver the highest return over the next 12 months, in terms of cost, revenue or satisfaction.
Service agents, sales agents, internal knowledge agents, and call quality assurance agents, built on Claude, GPT, Gemini or local models according to sensitivity.
Connecting AI to your CRM and existing systems, including Zapier, Make and n8n, in deployments that produce measurable savings.
A half-day or full-day in-house workshop that moves your team from basic ChatGPT use to professional work with models, prompts and context.
Evaluating three to five vendors against a real requirement, with no bias toward any particular supplier.
What ChatGPT, Gemini, Claude and Perplexity answer when someone asks about you, the other side of the same discipline, linked to the GEO page.
In a supervised organization, the AI question stops being a technology question the moment an agent gains access to customer data, and it becomes a governance question: who approves, who documents, and who is accountable when the system is wrong. Organizations that do not build that framework before deployment build it after the incident.
Not all data can leave the organization, so we map sensitivity levels and match the architecture accordingly, including local models or private deployment where required.
We define explicitly which decisions an agent may take alone, which require human approval, and what escalation to a representative looks like.
Every agent action touching a customer must be reconstructable after the fact, and that is an operational requirement before it is a regulatory one.
Amendment 13 to the Protection of Privacy Law, and sector guidance on outsourcing and cloud, affect the order of operations. They leave room for AI agents, and they determine what gets built first.
AI is new, and for us it is the third round of exactly the work Nekuda Digital has been doing since 1999, understanding how engines that rank content decide what to show. First it was AltaVista, then Google, and now answer engines.
Gilad Sasson runs the work personally on every project. He conducts the interviews, writes the document, sits in the execution meetings and holds quality from the kickoff to the end.
Guidance on selecting a first use case and building a governance framework before deployment, in an environment where data sensitivity dictates the architecture.
Process mapping and a review of existing systems to identify automation opportunities, under transparency and accessibility requirements.
Specification for integrating AI into internal search and personalization, across several languages and a large catalog.
We publish prices because an organization planning a budget deserves to know the order of magnitude before the first call.
| Track | For | Price | Duration |
|---|---|---|---|
| AI Readiness assessment | The entry point, any organization starting out | 9,900 ILS | 14 days |
| Work plan and deployment | After assessment, for a defined project | 35,000 to 90,000 ILS | 12 weeks |
| Monitoring and guidance subscription | Organizations with live systems needing ongoing measurement | From 5,700 ILS per month | 6 month minimum |
| Enterprise | Enterprises and institutions with multiple units and SLA requirements | From 18,000 ILS per month | Annual engagement |
What the monitoring subscription includes. A monthly run of the question set we started with across ChatGPT, Gemini, Claude and Perplexity, a live dashboard showing the trend over time, a biweekly review call, and ongoing adjustments to systems already in production.
Why a 6 month minimum. It comes from measurement, since under six months there are not enough data points to separate real improvement from noise, and we do not want to sell you a report you cannot draw conclusions from.
The Enterprise track includes all of the above, plus response time SLAs, a quarterly board-level review, parallel support for several business units, and support for procurement processes and institutional information security requirements.
Every final quote follows an introductory call, because scope in a large organization varies substantially with the number of systems and units involved.
Assessment within 14 days, a first pilot running within 30 to 60 days, measurable value within 90 days. We do not put clients into projects that run a year before producing a result.
The AI Readiness assessment starts at 9,900 ILS, a full 12-week work plan runs between 35,000 and 90,000 ILS depending on scope, the monitoring and guidance subscription starts at 5,700 ILS per month on a minimum six month engagement, and the Enterprise track starts at 18,000 ILS per month.
It is the first consideration, which is why it is examined during the assessment. In most large organizations the real constraint is which core systems are genuinely reachable via API, with model selection coming second, and sometimes the conclusion is that the first step is integration rather than AI, which we will say plainly.
Not necessarily, and this is determined by data classification. In sensitive cases we work with local models or private deployment, and data classification and architecture selection are a standard part of the assessment.
No. Every document and recommendation is written in plain language without jargon, and we work with CEOs, operations leads and IT managers. If something is unclear, simplifying it is our responsibility.
New agencies are excellent at technical deployment, and we bring the same deployment plus 27 years of understanding algorithms. Where AI touches marketing, search, or how your organization is represented in answer engines, that understanding is critical, and where the work is purely operational automation we sometimes recommend a smaller and cheaper partner.
Vendor agnostic: OpenAI, Anthropic, Google, Mistral and Cohere. Selection depends on the case, the budget and privacy requirements.
Yes. We know the requirements of public and institutional bodies and provide the necessary documentation, including information security and regulatory compliance responses.
Yes. Half-day and full-day in-house workshops are a core part of our work, and a workshop can be attached to a deployment project.
An initial conversation mapping what deserves investment at your organization this year and what does not. If it fits, we continue to a full assessment, and if not, you leave with two practical ideas for next week.
