Panel rating · 6 judges · How to read the stars

Category median

Sovereignty: not determined

0–5 in half steps. 5 means the rubric's top anchor is met on the evidence.

by OpenDialog AI Limited · www.opendialog.ai

Report an error on this page Is this your product? →

Link to this page

OpenDialog: AI-assisted analysis on whats-best.ai OpenDialog: AI-assisted analysis on whats-best.ai (dark)

Add ?theme=dark to the image address for dark backgrounds. The badge shows the category and the month of the latest evaluation, and EU-Made where that is the product’s provenance. It never shows a score.

OpenDialog at a glance

Vendor
OpenDialog AI Limited
Headquarters
Not stated on the vendor’s pages as of 17 Nov 2026
Data hosting
Not stated on the vendor’s pages as of 17 Nov 2026
Free plan
Yes, per the vendor’s own page source, captured 17 Nov 2026
Panel verdict
The AI panel rates OpenDialog 1.5/5, the mean of 6 judges. OpenDialog AI Limited sells named AI agents for insurance journeys — Selma for quote conversions, Jamie for policy questions, Michaela for policy changes, Rhea for re-engagement, Clark for claims — backed by the SAFER framework, a five-metric benchmark, continuous monitoring and a refusal test where an agent declined 220 of 222 questions on insufficient data. how scores are made
As of
Vendor pages captured 17 Nov 2026; evaluated by the panel 18 Nov 2026. all sources

OpenDialog alternatives, European ones first →

Read this page as one judge. Each weighs the same scores by what they care about.

The panel's verdict

OpenDialog AI Limited sells named AI agents for insurance journeys — Selma for quote conversions, Jamie for policy questions, Michaela for policy changes, Rhea for re-engagement, Clark for claims — backed by the SAFER framework, a five-metric benchmark, continuous monitoring and a refusal test where an agent declined 220 of 222 questions on insufficient data. Agent capabilities are the strength: five of six judges scored 6, the skeptic 0, citing no public information on human approval before consequential actions, triggers and schedules, or per-run logs. Around the agents the pages go quiet. Processing and models, sovereignty and pricing transparency each sit at 0: no model, provider, hosting location or subprocessor is named on the pages, and no prices are published — the route in is a demo request and a free whitepaper. Judges split on admin and access, 0 to 4, and integrations and knowledge, 0 to 3 — weighing a documented audit trail, analytics and published retrieval figures against no public information on SSO, roles, connectors or APIs.

Report an error

Speaks for it

  • Five named agents cover quotations, policy questions, policy changes, re-engagement and claims, with custom agents built on request.
  • Published safety evidence includes a five-metric SAFER benchmark, continuous monitoring and a refusal test where an agent declined 220 of 222 questions on insufficient data.
  • Retrieval answers from company knowledge, with 95% retrieval precision and recall reported and access to knowledge sources required at onboarding.
  • A detailed audit trail and extensive analytics are documented, with admin control over dynamic versus static content and knowledge retrieval.
  • Agent capabilities drew 6 from five of six judges, the highest mark any criterion received.

Report an error

Held against it

  • No public prices of any kind — no tiers, seats or usage figures — so an annual cost cannot be computed from public pages.
  • No public information on which models run the agents, who operates them, or where inference and storage happen.
  • No sovereignty attributes on record, with no public information on hosting location, the contracting entity's jurisdiction or subprocessors including model providers.
  • No public information on single sign-on, user provisioning, roles or group permissions.
  • No public information on whether prompts or outputs train a model, on retention periods, or on a data processing agreement.

Report an error

Best for

  • You run insurance journeys — quotations, policy questions, policy changes, re-engagement, claims — and want named agents purpose-built for those steps.
  • You want published evaluation evidence before go-live: a five-metric benchmark, continuous monitoring and demonstrated refusal behaviour.
  • You operate under regulatory mapping such as FCA guidelines and need a detailed audit trail and extensive analytics.
  • You are prepared to start with a demo request and a 3-hour discovery workshop.

Report an error

Avoid if

  • You must forecast spend before talking to sales, and no public price exists to compute an annual cost.
  • Your procurement requires named model providers, subprocessors and hosting location before signature.
  • You need single sign-on, user provisioning and role-based permissions documented before a large rollout.
  • You require a documented human approval step before agents change policies or process claims.

Report an error

The scores

Panel median The median of the judges’ scores per criterion. Each label opens its reasoning below.
Scores per criterion, in stars out of 5
Criterion Panel median
Data use, training & retention 0.5 of 5
Admin, identity & access controls 1.5 of 5
Integrations, connectors & knowledge 1.5 of 5
Agent capabilities & guardrails 3 of 5
European sovereignty 0 of 5
Pricing transparency 0 of 5

Data use, training & retention

n/r Show reasoning
How this is scored

What happens to prompts, uploaded files, connected data and outputs: whether any of it trains a model (the vendor's or a model provider's), how long it is kept and where logs live, whether zero-retention is available, and whether a data processing agreement covers it.

0 — No statement on training or retention; the privacy policy is a website policy that does not cover the product.

3 — A general statement that business data is "not used for training", with no word on retention, on the model providers behind the product or on logs.

5 — Business data excluded from training by contract, retention periods stated for chats and files, and a DPA offered to business customers.

8 — No-training commitments stated for the vendor and every model provider it routes to, retention configurable by the admin (including deletion of chat history), zero-data-retention terms with the model providers named, and abuse-monitoring retention described.

10 — Data use fully accounted for: training, retention, logging and abuse monitoring stated per model provider, customer-controlled retention and deletion with an audit trail, outputs and connected data treated as customer data in the DPA, and an independent audit (SOC 2, ISO 27001 or ISO 42001) covering the product.

Report an error

The Head of IT

The captured pages offer only links to terms of use and a privacy policy; we found no public information on whether prompts or connected insurance data train a model, on retention periods or zero-retention terms, on the model providers behind the product, or on a data processing agreement. 1

Report an error

The Data Protection Officer

The pages confirm only that a privacy policy and terms of use exist; we found no public information on whether prompts, claims files or outputs train any model, on retention periods for chats or files, on abuse-monitoring logs, or on a data processing agreement for business customers. For a product that touches insurance claims and policy data, I would not sign a DPA on this basis. 1

Report an error

The Automation Lead

Terms of use and a privacy policy are named and a detailed audit trail is claimed, but I found no public information on whether prompts or outputs train a model, how long conversations and files are kept, zero-retention terms, or a data processing agreement. For agents that handle claims and policy changes, that silence is the first thing I would push on. 1 3

Report an error

The Customer-Service Director not rated — no public information

We found no public information on whether customer conversations are used to train models, how long chats and files are kept, or whether a data processing agreement is offered; the captured pages point to terms of use and a privacy policy without stating what they cover. Nothing here tells me what happens to our claims chats and transcripts. 1 2 3

Report an error

The Works Council Member

The captured pages tell us only that a terms of use and a privacy policy exist, and we found no public information on whether prompts or customer data train a model, on retention periods, on logs, or on a data processing agreement. Before any assistant touches staff or customers I need prompt logging rules set out in writing, and nothing captured describes them. 1

Report an error

The Skeptic

Beyond the listed titles "Terms of use" and "Privacy policy", we found no public information on whether prompts, files or outputs train a model, how long anything is kept, or whether a data processing agreement is offered. 1

Report an error

Models & processing location

n/r n/r n/r n/r n/r n/r Show reasoning
How this is scored

Which models sit behind the product and who runs them, where inference and storage happen, whether the customer can pin processing to the EU, choose or exclude model providers, bring its own model or key, or run the product in its own cloud or on premises.

0 — The models behind the product and the processing location are not stated.

3 — The model family is named, but where inference runs and through which provider (the model maker, a hyperscaler, the vendor) is unstated.

5 — Models and their providers named, with an EU processing option stated for storage and at least the default model.

8 — Each available model listed with provider and inference region, EU-only processing selectable by the admin and enforced across models, model providers excludable, and bring-your-own-key or own-cloud deployment documented.

10 — Processing fully under the customer's control and documented: EU-only or on-premises inference for every feature including agents and connectors, open-weight or self-hosted models supported, data flows per feature published, and a residency commitment in the contract rather than on a marketing page.

Report an error

The Head of IT not rated — no public information

We found no public information on which models sit behind the named agents, who runs them, or where inference and storage take place, let alone EU pinning, provider exclusion or bring-your-own-key deployment. 1 2 3

Report an error

The Data Protection Officer not rated — no public information

We found no public information on which models sit behind these agents, who operates them, or in which region inference and storage happen — the homepage, the agents page and the safety article name no model or provider. I cannot assess an EU-only processing option that covers agents and connectors because the captured pages describe none. 1 2 3

Report an error

The Automation Lead not rated — no public information

I found no public information on which models sit behind these agents, who runs inference, or where it runs. Neither the homepage nor the agents page names a model family, a provider, an inference region, an EU pinning option, or a bring-your-own-key or self-hosted route. 1 2

Report an error

The Customer-Service Director not rated — no public information

We found no public information on which models sit behind the agents, who runs them, or where inference and storage happen, so I cannot tell whether our conversations would ever be processed outside the EU. 1 2 3

Report an error

The Works Council Member not rated — no public information

We found no public information on which models sit behind the agents, who runs them, or where inference and storage happen; no model provider or processing region is named on any captured page. Without that, no co-determination over data flows is even possible. 1 2 3

Report an error

The Skeptic not rated — no public information

Which model answers, who runs it and where inference happens are the first questions I ask, and we found no public information on any of them: the captured pages name no model, no provider and no processing region. 1 2 3

Report an error

Admin, identity & access controls

panel disagrees Show reasoning
How this is scored

What the IT department controls: SSO and SCIM provisioning, roles and groups, which models, agents and connectors each group may use, retrieval that respects the permissions of the source system, audit logs, usage reporting, and policies such as data-loss prevention — and whether reporting on staff use is scoped and documented.

0 — No admin features documented; accounts are individual sign-ups.

3 — A team workspace with an owner and member invitations, but no SSO, no roles and no audit log.

5 — SSO, admin and member roles, user management, and an admin view of usage.

8 — SAML/OIDC SSO with SCIM provisioning, group-based permissions for models, agents and connectors, retrieval that inherits source-system permissions, exportable audit logs, and usage analytics with documented scope.

10 — Governance a regulated company can operate: fine-grained policies (DLP, content filters, prompt and output logging rules), audit logs streamed to a SIEM, per-agent ownership and approval, usage reporting that can be aggregated or pseudonymised for works-council agreements, and every admin action logged.

Report an error

The Head of IT

A detailed audit trail, extensive analytics and ongoing monitoring are claimed for the agents, but we found no public information on SSO, SCIM provisioning, roles and groups, or retrieval that respects source-system permissions — the controls my eight-hundred-staff rollout depends on. 3

Report an error

The Data Protection Officer

The vendor commits to "a detailed audit trail and extensive analytics" and to controls over knowledge retrieval versus static content, which an IT department can at least point at. We found no public information on single sign-on, user provisioning, roles, or group-level permissions over models, agents and connectors. 3

Report an error

The Automation Lead

A detailed audit trail, extensive analytics, continuous monitoring and admin control over dynamic versus static generation and knowledge retrieval are all claimed. But we found no public information on single sign-on, provisioning, roles or group-level permissions over which teams can use which agents, which is what I need before pointing this at an ERP. 3

Report an error

The Customer-Service Director

A detailed audit trail and extensive analytics are documented, together with control over dynamic versus static content and knowledge retrieval, and mapping to regulatory requirements such as FCA guidelines. What my IT team needs most — SSO, roles, provisioning, group permissions — is undocumented on the captured pages. 3

Report an error

The Works Council Member

A detailed audit trail and extensive analytics are promised, with agents said to run under continuous monitoring and to be auditable against regulatory expectations — a basis for oversight, at least as a claim. But we found no public information on single sign-on, roles or group permissions, and nothing on the scope of usage reporting, which for a works-council agreement must be aggregable or pseudonymised. 1 3

Report an error

The Skeptic

The captured pages promise a "detailed audit trail and extensive analytics" and a control between dynamic and static content or knowledge retrieval, but we found no public information on single sign-on, user provisioning, roles, or which teams may use which agent. 1 3

Report an error

Integrations, connectors & knowledge

panel disagrees Show reasoning
How this is scored

How the product reaches company knowledge and systems: connectors to file stores, wikis, mail, CRM, ticketing and ERP, how fast and how completely they sync, the API, support for open protocols such as MCP, and how the assistant cites the sources it answers from.

0 — No integrations; users paste text or upload files one at a time.

3 — File upload and one or two connectors (for example a drive), with no API and no statement on sync or citations.

5 — A connector library covering the common file stores, wikis and collaboration tools, answers citing sources, and a documented API.

8 — Connectors to file stores, wikis, mail, CRM and ticketing with incremental sync, custom connectors or MCP support, an API and webhooks for embedding assistants and agents in other tools, and source citations on every grounded answer.

10 — The product sits in the company's systems: read and write connectors with documented scopes, MCP servers and clients, contact-centre or telephony integration where relevant, a knowledge layer that can be audited and corrected, and the data, agents and configuration exportable when the customer leaves.

Report an error

The Head of IT

The agents demonstrably answer from company knowledge — retrieval precision and recall of 95 percent is reported and onboarding requires access to knowledge sources — but we found no public information on which connectors exist, an API or webhooks, sync speed and coverage, or how answers cite their sources. 1 3

Report an error

The Data Protection Officer

The agents demonstrably answer from customer knowledge — onboarding requires "relevant access to knowledge sources" and the vendor publishes retrieval precision and recall figures — but we found no connector catalog for file stores, mail or policy systems, no documented API, and no statement on sync behaviour or how answers cite their sources. 1 3

Report an error

The Automation Lead

Knowledge retrieval is real enough to be measured: 95% retrieval precision and recall is quoted, onboarding asks for access to knowledge sources, and the framework controls whether agents retrieve or generate. We found no public information on connectors, a documented API, MCP support, sync behaviour, or how answers cite their sources. 1 3

Report an error

The Customer-Service Director

Discovery requires relevant access to our knowledge sources, and reported figures of 95% retrieval precision and recall plus an agent refusing 220 of 222 questions on insufficient data show a real grounding layer. We found no public information on a connector catalogue for our ticketing and CRM, an API, sync behaviour, or citations on answers. 1 3

Report an error

The Works Council Member

Deployment requires relevant access to knowledge sources, and the vendor publishes retrieval precision and refusal figures, so grounding in company knowledge is real. We found no public information on specific connectors, an API, sync behaviour, or how an answer shows the sources it drew from. 1 3

Report an error

The Skeptic

Retrieval is real — customers must give "relevant access to knowledge sources" and the vendor claims "95% retrieval precision/recall" — but we found no public information on named connectors, sync behaviour, an API, or how answers cite their sources. 1 3

Report an error

Agent capabilities & guardrails

panel disagrees Show reasoning
How this is scored

What an agent can actually do and where it stops: multi-step tasks, tool use and actions in other systems, scheduling and triggers, human approval before consequential actions, run logs, evaluation and testing before release, and the limits the vendor states rather than hides.

0 — No agents; the product is a chat window, whatever the marketing says.

3 — Saved prompts or custom assistants with instructions and files, but no actions in other systems and no run history.

5 — Agents that call tools or APIs and run multi-step tasks, built without code, with a log of each run.

8 — Agents that act in connected systems with scoped permissions, human approval configurable per action, triggers and schedules, test and evaluation tooling before release, and documented limits and failure handling.

10 — Agents a business can put in production and account for: versioning and staged rollout, evaluation against test sets, monitoring of quality and cost per agent, handover to a human with full context, every action attributable to the agent and the person who deployed it, and the vendor's own statement of what agents must not be trusted with.

Report an error

The Head of IT

Five named agents run end-to-end insurance journeys (quotations, policy questions, policy changes, re-engagement, claims) with an audit trail, and the SAFER framework adds evaluation across five metrics, continuous monitoring, and a published test in which an agent refused 220 of 222 questions when data was insufficient. We found no public information on human approval before consequential actions, triggers and schedules, or per-agent quality and cost monitoring in production. 1 2 3

Report an error

The Data Protection Officer

Five named agents cover quotations, policy questions, policy changes, re-engagement and claims, and the guardrail story is unusually concrete: a published safety benchmark on five metrics, continuous monitoring, an audit trail, and a test where an agent refused 220 of 222 questions it lacked data for. We found no public information on human approval before consequential actions, scoped per-agent permissions, or triggers and schedules, so I would not yet clear this for unattended claims handling. 1 2 3

Report an error

The Automation Lead

Five named agents do consequential work across quotations, policy questions, policy changes, claims and re-engagement, backed by an audit trail, continuous monitoring, a SAFER benchmark for evaluating agents before and during operation, and a documented refusal behaviour (220 of 222 questions refused when data was insufficient). What I could not find is human approval before an action fires, scoped permissions per agent, or triggers and schedules, which is precisely the guardrail set I insist on. 1 2 3

Report an error

The Customer-Service Director

Named agents handle quotations, policy questions, policy changes, re-engagement and claims, custom agents are built on request, and the SAFER framework with its five-metric benchmark and continuous monitoring gives me the evaluation-before-go-live I insist on, with documented refusal behaviour when knowledge is missing. What I could not find is configurable human approval before consequential actions or a handover to a human with full context — the make-or-break for my forty-seat centre. 1 2 3

Report an error

The Works Council Member

Named agents run whole insurance journeys — quotations, policy changes, claims, re-engagement — under a published safety framework with a five-metric evaluation benchmark, continuous monitoring, and demonstrated refusals when knowledge is insufficient, which is more honest about limits than most vendors manage. What we found no public information on is a human approval gate before consequential actions, scoped permissions for what an agent may touch, and triggers or schedules — and an agent that changes people's policies without a documented approval step is exactly what a works council cannot sign off. 1 2 3

Report an error

The Skeptic

Five named agents run multi-step insurance journeys from quotation to claims and policy changes, and the vendor publishes a five-metric SAFER benchmark, continuous monitoring, and a refusal test where an agent declined 220 of 222 questions beyond its knowledge. We found no public information on human approval before consequential actions, triggers and schedules, or per-run logs. 1 2 3

Report an error

European sovereignty panel opinion

n/r n/r n/r n/r n/r Show reasoning
How this is scored

Who the contracting entity and the parent company are, where customer data is hosted, and who the subprocessors are — model providers above all, then hosting, search, telephony and email. Independently sourced by the sovereignty pipeline; for an AI product the model provider is a subprocessor, so a European interface over a non-European model counts as such.

0 — Non-EU vendor and contracting entity, hosting unstated, subprocessors (including model providers) unnamed.

3 — A privacy policy exists, but hosting location is unstated or non-EU and the model providers behind the product are not named as subprocessors.

5 — EU hosting stated and an EU contracting entity, but prompts reach non-EU model providers or other subprocessors without a stated safeguard or EU region.

8 — EU hosting on named infrastructure, EU contracting entity, every subprocessor listed with its location including model providers, EU inference available, and a DPA offered.

10 — Sovereign end to end and evidenced: European ownership, entity, hosting and model processing, every subprocessor European or optional, data export and deletion documented, and a security or certification statement published rather than implied.

Report an error

The Head of IT not rated — no public information

We found no public information on hosting location, the contracting entity beyond the company name, or subprocessors including the model providers; the only geographic signal on the pages is a mapping to FCA guidelines. 1 3

Report an error

The Data Protection Officer not rated — no public information

We found no public information on hosting location, the contracting or parent entity, or a subprocessor list — and for an AI product the model provider is a subprocessor, so its name and region are the first thing I look for. The captured pages publish none of it, so no sovereignty dimension can be scored from public evidence. 1 2 3

Report an error

The Automation Lead not rated — no public information

No sovereignty attributes are on record and the captured pages say nothing about hosting location, the contracting entity's base, or subprocessors, including which model providers receive the prompts. I found no public information on any European hosting or inference commitment, so there is nothing here for me to score. 1 2 3

Report an error

The Customer-Service Director not rated — no public information

We found no public information on where customer data is hosted, where the contracting entity sits, or which subprocessors — above all the model providers — are involved, so I can confirm no European processing or hosting commitment from these pages. 1 2 3

Report an error

The Works Council Member not rated — no public information

We found no public information on where customer data is hosted, on the contracting entity beyond the vendor name, or on subprocessors — for an AI product the model provider is the first subprocessor I would need named, and none is. Every dimension here is unknown rather than answered. 1 2 3

Report an error

The Skeptic

Terms and a privacy policy are listed, but we found no public information on hosting location, the contracting entity's jurisdiction, or any subprocessor — including the model providers every prompt must reach. 1 2 3

Report an error

Pricing transparency

n/r Show reasoning
How this is scored

Whether a company can compute the real annual cost from public pages alone: price per seat, per agent, per conversation or per usage unit, what is included in each tier, usage limits and overage, which models, connectors or agents cost extra, minimum seats and term.

0 — No public prices at all; every customer is a sales call.

3 — A starting price or "from" figure exists, but usage limits, included models and what scales the cost are unclear.

5 — Prices per tier are public with the main inclusions, but usage limits, overage, agent runs or enterprise features (SSO, audit logs, EU processing) are unpriced or "on request".

8 — Every tier priced publicly with seat or usage units, usage limits and overage rates, add-ons listed, minimum seats and term stated.

10 — Complete price computability: annual cost derivable for a given number of seats, agents or conversations, with every usage rate, model surcharge, add-on and support tier and the cancellation and data-return terms published.

Report an error

The Head of IT

The captured pages show no prices of any kind; the stated route in is a form to request a demo and a whitepaper you can download for free, so an annual cost cannot be computed from public pages. 2 3

Report an error

The Data Protection Officer

No seat, agent or usage price is published on the captured pages — the public path is "Request a demo" plus a free whitepaper download, which is documentation rather than pricing. A company cannot compute any annual cost from this material. 2 3

Report an error

The Automation Lead

The only priced item in the captures is a whitepaper offered free ('Complete the form and download instantly for free'), sitting next to a demo-request form. No seat, tier, agent or usage prices are published, so no annual cost can be computed from public pages. 2 3

Report an error

The Customer-Service Director not rated — no public information

We found no public pricing information — nothing per seat, per agent or per conversation — only a free whitepaper to download instantly and a demo request. There is nothing here my CFO could forecast a per-conversation cost from. 2 3

Report an error

The Works Council Member

We found no public information on prices for any tier, per seat or per usage; the documented buying route is a demo request and custom agents built on request, with a whitepaper downloadable for free as the only figure published. No company can compute an annual cost from these pages. 1 2 3

Report an error

The Skeptic

The only cost-adjacent items on the captured pages are a free whitepaper download ("Complete the form and download instantly for free") and a demo request form; we found no public prices, tiers, seat counts or usage limits, so an annual cost cannot be computed from these pages. 2 3

Report an error

Questions about OpenDialog

Is OpenDialog European?

Not determined: the vendor’s pages as captured do not say where OpenDialog is based.

Where does OpenDialog store data?

Not stated on the vendor’s pages as of 17 Nov 2026.

Does OpenDialog have a free plan?

Yes. OpenDialog has a free plan, per the vendor’s page captured 17 Nov 2026.

What are the best alternatives to OpenDialog?

As of 11 Nov 2026, the panel rates Langdock (4/5), Dust (3.5/5) and Unique AI (3.5/5) highest among the other 23 AI Agents & Assistants products evaluated. Ratings that look equal are rounded to half stars; the order uses the unrounded rating, then the name. The highest-rated European alternative is Langdock (EU-Made, Germany). Ratings are the AI panel’s opinion, formed from public evidence: how scores are made. The full list, with European products first: OpenDialog alternatives.

Where this could be wrong

What we left out

A claim that does not survive our checks costs us the claim, not the page. This is what was taken off this one.

Sources (3)

The pages every claim on this page was read from — each one checked, dated, and kept verifiable.

  1. 1 Vendor page www.opendialog.ai Checked 17 Nov 2026 Details →
  2. 2 Agent capabilities & guardrails — found from sitemap www.opendialog.ai Checked 17 Nov 2026 Details →
  3. 3 Agent capabilities & guardrails — found from sitemap www.opendialog.ai Checked 17 Nov 2026 Details →