DataWiseStart a scoping conversationScoping callעברית
Menu

AI Agent Companies in Israel: Wonderful and the Alternatives

AI agent companies in Israel come in four types: agent platforms sold with their own deployment team, such as Wonderful and Notch; enterprise integrators; bot and automation shops for small businesses; and hands-on engineering teams that build the agent into your existing systems, such as DataWise. Which fits depends on your size, systems, Hebrew and who owns the result.

Last updated 2026-10-20. This is an interim guide, and we update it as the market moves. Every fact about another company links to its source; all sources were read on 2026-10-19 and are listed at the end.

Wonderful (wonderful.ai) is an Israeli-Dutch AI agent company founded in early 2025 (TechCrunch, 2026-09-02), which announced a $550 million Series C at a $5 billion valuation on 2026-09-02 (Wonderful). It sells AI agents together with the engineers who deploy them inside the customer’s organization, and about 400 of its 650 employees work at customer sites (Calcalist, 2026-09-06).

It is built for large, regulated organizations. Its home page reads “Applied AI for the enterprise” (wonderful.ai), its industry pages cover banking, insurance, healthcare, telecom, retail and utilities (banking, for example), and the three case studies on its site come from an Israeli bank, health fund and telecom operator (financial services, healthcare, telecom).

This guide is for a different buyer: a mid-sized company in Israel looking for an AI agent company, and weighing a platform like Wonderful against a team that builds the agent into the systems it already runs, in Hebrew.

Four types of AI agent supplier

Agent platforms with forward-deployed teams

A software platform for building and running agents, sold together with engineers who deploy it inside the customer’s organization. Wonderful works this way. Its Forward Deployed Engineers build “the agents, integrations, and workflows into your stack” and work on site, alongside Deployment Strategists who define the use cases and metrics with the customer (deployment page). Besides customer service, its industry pages describe back-office work such as KYC, reconciliation and case management (banking).

The model is not unique to Wonderful. Sierra offers to “work with an expert agent development team that has supported hundreds of deployments” (Sierra). Decagon credits “Forward-Deployed Engineering Support” in its Rippling case study (Decagon). Notch, an Israeli company (Globes, 2026-03-25) with agents for insurance operations (Notch), says “a dedicated implementation team handles configuration, integration, and testing” (Notch). In each case the software comes with the people who deploy it.

Enterprise integrators

Large consultancies and systems integrators that run organization-wide programs: change management, procurement frameworks, a program office. They and the platforms can work together: Wonderful’s press page lists a partnership with McKinsey, dated 2026-04-06 (Wonderful press). This is the right column for a rollout across thousands of employees, and usually too much structure for one operational problem in a mid-sized company. Three ways to buy AI covers this column in more detail.

Bot and automation shops for small businesses

Agencies and consultants that sell WhatsApp and website bots and workflow automations to small businesses. They fit a self-contained task with a low cost of error: answering common questions, booking appointments, capturing leads. Meta now offers one of its own: it made its AI agent for WhatsApp Business available globally on 2026-06-03, and it answers questions, books appointments, qualifies leads and hands off to a person (TechCrunch, 2026-06-03). This type strains when the agent has to read and write your ERP, or act on money.

A hands-on engineering team

A small senior team that takes one operational problem, builds the agent into the systems you already run, and hands it over. This is where DataWise sits. The same senior engineer who characterizes the problem builds the solution and ships it into your systems (how we work). We have scoped and built against Priority, SAP, monday.com and Hashavshevet environments (AI-ready data), and the operations agent we built ran in WhatsApp and took Hebrew voice messages as well as text (AI agents). You own the models, the code and the pipelines, with no platform meter.

Agents are one of the things we build, next to forecasting, computer vision and AI for closed networks. The limit is scale: a small team cannot run a contact center for millions of customers or an organization-wide program, and should not pretend to.

When a platform like Wonderful is the right call

A platform with a forward-deployed team fits when one or more of these is true:

  • Customer conversations at national scale. Wonderful’s case studies are about customer service by voice at large organizations: appointment booking at a bank and at a health fund, technician scheduling at a telecom operator (financial services, healthcare, telecom).
  • An organization-wide program, sponsored from the top. Wonderful’s own essays address the CEO: “It has to be sponsored at the CEO level” (Wonderful).
  • Your own engineers will keep building agents. In Wonderful’s Bank Hapoalim story, a data engineer at the bank with no prior experience building agents shipped a new agent in three weeks, reusing 15 tools that Wonderful’s engineers had built (Wonderful).
  • Procurement requires vendor certifications. Wonderful lists SOC 2 Type II, ISO 27001:2022, GDPR and PCI DSS (AI OS page).

If several of these describe you, start with the platforms. If you are a mid-sized company with one operational problem, Hebrew material and an ERP such as Priority, the table below sets out the choice.

The decision table

The two delivery models side by side, with the question worth asking either one.

Criterion Agent platform with a forward-deployed team Hands-on engineering team What to ask either one
Organization size Large and regulated. Wonderful’s home page reads “Applied AI for the enterprise” and “Deep in regulated, mission-critical industries” (wonderful.ai). Small and mid-sized. DataWise is sized for Israeli SMB and mid-market operations, not hundred-seat enterprise programs (transformation). Which of your customers are our size?
Systems: Priority, SAP, monday Built to connect to “the systems the work runs on” (agents); Wonderful’s computer-use agents also work in legacy systems (computer use). We have scoped and built against Priority, SAP, monday.com and Hashavshevet (FAQ). Which of our systems have you connected before, by name?
Hebrew Agents in local languages. Wonderful’s Maccabi story lists Hebrew, English, Arabic, Russian and French (Maccabi). Our WhatsApp operations agent took Hebrew voice messages as well as text (AI agents). Can we see it handle our own Hebrew material, voice notes included?
Closed network or data residency Offered. Wonderful lists multi-tenant, single-tenant, bring-your-own-cloud and on-premise air-gapped deployment (deployment). Offered. Self-hosted models, retrieval and vision inside closed networks, with nothing leaving them (secure AI). Where do the data and the model run, and what leaves our network?
Who owns the code Ownership passes to you by design, “from Wonderful-led to full client ownership” (deployment). Wonderful describes bringing together “the AI platform, deployment teams, and strategic partnership” (wonderful.ai). You own the models, the code and the pipelines. No platform meter, no lock-in (how we work). What keeps running if we part ways in a year?
Cost structure Not published. Wonderful’s site routes buyers to “Set a meeting with our team” (contact); Sierra describes outcome-based pricing without rates (Sierra). Hourly advice, a fixed price per stage, then a monthly retainer. Scoping is a fixed price quoted after a short call (scoping price). What does the first stage cost, and what do we keep if we stop there?
Who builds and who runs it The vendor’s engineers build on site, with a strategist who sets the use cases and metrics (deployment). The engineer who scopes it builds it. Afterwards your team runs it, or a monthly retainer covers running it (scoping price). Will the person who scopes this also build it?

For the same map across all of AI rather than agents alone, see three ways to buy AI. For the questions to put to any supplier, us included, see questions worth asking before hiring.

How we deploy

The engineer who scopes the problem builds it and ships it into your systems, in three stages with an exit point after each: characterization, feasibility on your real data, then implementation and handover (how we work). We are on site when the work needs it (kickoff, the scoping sample, go-live), and remote otherwise. What the agent does alone and what waits for a person is set in writing at characterization (permissions and boundaries). What drives the price, and why a number before scoping is a guess, is on the pricing page.

Find the page for your problem

Sources

All read on 2026-10-19. Publication dates are given where the source states one.

What does Wonderful AI do?

Wonderful builds AI agents for large organizations and sells them together with the engineers who deploy them inside the customer's business, a model it calls forward-deployed. Its agents work across voice, chat and email, mostly in customer service, and its industry pages also describe back-office workflows in banking, insurance, healthcare, telecom, retail and utilities. Each fact on this page links to its source and the date it was read.

What are the alternatives to Wonderful AI for a mid-sized company in Israel?

Four types of supplier sell AI agents in Israel: agent platforms with forward-deployed teams, such as Wonderful; enterprise integrators; bot and automation shops for small businesses; and hands-on engineering teams that build the agent into the systems you already run. Which one fits depends on your size, your systems, whether the work is in Hebrew, where the data may sit, and who should own the result.

When is a platform like Wonderful the right choice?

When customer conversations run at national scale, when the program is organization-wide and sponsored from the top, when your own engineers will keep building agents on a platform, or when procurement requires vendor certifications. If none of these describes you, a hands-on engineering team is worth weighing.

Does DataWise build AI agents that work in Hebrew?

Yes. The operations agent we built ran in WhatsApp and took Hebrew voice messages as well as text, and we have scoped and built against Priority, SAP, monday.com and Hashavshevet environments. Agents are one of several things we build, alongside forecasting, computer vision and AI that runs inside closed networks.

Start a scoping conversation