AI Agents & Assistants
Ada
Provenance unknown Report an error0–5 in half steps. 5 means the rubric's top anchor is met on the evidence.
by Ada Support Inc. · ada.cx
Report an error on this page Is this your product? →
Link to this page
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.
Ada at a glance
- Vendor
- Ada Support Inc.
- Headquarters
- Not stated on the vendor’s pages as of 3 Nov 2026
- Data hosting
- Not stated on the vendor’s pages as of 3 Nov 2026
- Free plan
- Not stated on the vendor’s pages as of 3 Nov 2026
- Panel verdict
- The AI panel rates Ada 2/5, the mean of 6 judges. Ada is an AI agent platform whose documentation is strongest on connections and controls. how scores are made
- As of
- Vendor pages captured 3 Nov 2026; evaluated by the panel 3 Nov 2026. all sources
Read this page as one judge. Each weighs the same scores by what they care about.
The panel's verdict
Ada is an AI agent platform whose documentation is strongest on connections and controls. Integrations and knowledge scored 6 to 7, built on eight knowledge-source connectors — Zendesk, Salesforce, ServiceNow and Microsoft Dynamics among them — syncing as fast as every 15 minutes; admin and access also scored 6 to 7, on SAML 2.0 single sign-on and a filterable, CSV-exportable audit log; agent capabilities holds at 6. The weakness sits underneath. Processing and models scored 1 to 2: no captured page names the models behind the product, who runs them or where inference happens. Sovereignty scored 1 to 2 — we found no public information on the contracting entity's jurisdiction, data residency, hosting or subprocessors; the only regional signal is US, EU and CA (Maple) agent endpoints. Data use and training holds at 4: retention is concrete — attachments after 7 days, end-user data after 2 years — but we found no public information on whether data trains models or on a data processing agreement. Judges' scores sit close together, with no criterion spanning more than one point; no prices appear on any captured page.
Speaks for it
- Retention documented in concrete figures — attachments deleted after 7 days, end-user data after 2 years — with a bulk deletion API, GDPR erasure on demand and export before expiry
- Eight knowledge-source connectors documented, including Zendesk, Salesforce, ServiceNow and Microsoft Dynamics, syncing as fast as every 15 minutes
- SAML 2.0 single sign-on with any identity provider, plus an audit log filterable by actor, activity, entity and interface and exportable as CSV
- Every Action run logs the request data, response data and whether it succeeded, with sample-value testing plus test conversations, test cases and simulations
- Playbooks combine conditions, actions and responses into multi-step workflows built without code
Held against it
- No captured page names the models behind the product, who runs them or where inference happens — processing and models scored 1 to 2
- Sovereignty scored 1 to 2 — we found no public information on the contracting entity's jurisdiction, data residency, hosting or subprocessors, and the only regional signal is US, EU and CA (Maple) agent endpoints
- We found no public information on whether prompts, transcripts or knowledge train a model, or on a data processing agreement
- Provisioning is manual — accounts are deleted and re-added to move users onto single sign-on, and initial setup requires an Ada representative
- We found no public information on configurable human approval before consequential agent actions, or on triggers and schedules
Best for
- You run customer support on Zendesk, Salesforce, ServiceNow or Microsoft Dynamics and want your existing knowledge base to drive the agent
- Your IT team needs SAML 2.0 single sign-on and a searchable audit trail of who changed what, when and through which interface
- You need published retention schedules backed by deletion tooling — bulk deletion API, GDPR erasure on demand, export before expiry
- You want multi-step workflows built without code, with each run's request and response data logged
Avoid if
- You must know which models run your conversations and where inference happens before procurement — processing and models scored 1 to 2
- You cannot sign without a data processing agreement and subprocessor disclosure — sovereignty scored 1 to 2 and we found no public information on either
- You expect automated user provisioning — moving accounts onto single sign-on is documented as a manual delete-and-re-add, and initial setup runs through an Ada representative
- Your knowledge lives in file stores or mailboxes rather than help centres — the documented connectors cover knowledge bases and service desks
The scores
| Criterion | Panel median |
|---|---|
| Data use, training & retention | 2 of 5 |
| Models & processing location | 1 of 5 |
| Admin, identity & access controls | 3 of 5 |
| Integrations, connectors & knowledge | 3.5 of 5 |
| Agent capabilities & guardrails | 3 of 5 |
| European sovereignty | 1 of 5 |
Data use, training & retention
Show reasoningHide 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.
The Head of IT
Retention is documented in real detail: end-user data and transcripts deleted after two years, attachments after seven days, dashboard visibility with archiving, on-demand GDPR erasure, a bulk deletion API, export before expiry, PII redaction, and a zero-data-retention statement for LLM processing. I found no public information on whether prompts or transcripts train any model, on who the model providers are, or on a data processing agreement — for an 800-seat rollout that gap decides the score. 1 2
The Data Protection Officer
Retention is properly published: end-user data deleted after two years, attachments after seven days, transcripts deleted after two years unless personal details are removed, with a bulk deletion API, on-demand GDPR erasure and export before deletion — and LLM calls are described as zero-data-retention, processed in memory for the duration of the request. What I cannot find is any statement on whether prompts, transcripts or connected knowledge are used to train a model, the identity of the model providers behind that zero-retention processing, or a data processing agreement covering all of it. As it stands I could not sign this into a DPA. 1 2
The Automation Lead
Retention is documented in real detail: end-user data auto-deleted after two years, attachments after seven days, conversations archived after roughly six months of dashboard visibility, plus a bulk deletion API, on-demand GDPR erasure, redaction settings, and in-memory zero-data-retention handling for model processing. I found no public information on whether prompts or outputs train any model, on a data processing agreement, or on the model providers behind the product, which keeps this short of a clean bill. 1 2
The Customer-Service Director
Retention is documented in real numbers — attachments go after seven days, end-user data and transcripts after two years — with a bulk deletion API, GDPR erasure on demand, redaction, and the language-model layer described as in-memory only under zero-data-retention principles. I found no public information on whether chats or knowledge are used for training, or on a data processing agreement, which is the gap my privacy officer will chase before go-live. 1 2
The Works Council Member
Retention is documented in detail: attachments go after seven days, end-user data and transcripts after two years, LLM requests are processed in memory under zero-data-retention principles, and there is a bulk deletion API, GDPR erasure on demand and pre-storage redaction of personal details. What is missing completely is the works-council question — I found no public information on whether prompts, transcripts or knowledge ever train a model, and no data processing agreement is mentioned on any page I read. 1 2
The Skeptic
Retention is genuinely documented — attached files go after seven days, end-user data after two years, transcripts with personal data likewise, with a bulk deletion API, export before expiry and on-demand GDPR erasure — and the vendor states model processing follows zero-data-retention principles, held in memory only for the duration of the request. But no captured page states whether anything trains on customer data, names the model providers behind the product, or offers a data processing agreement, and retention is described as a fixed schedule rather than something an admin configures. I cannot sign off on a data posture when I do not know who the other processors are. 1 2
Models & processing location
Show reasoningHide 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.
The Head of IT
I found no public information on which models sit behind the product, who runs inference, or in which region; the only location signal anywhere is that instances exist in US, EU and CA (Maple) regions, seen in a connector webhook setup page. No EU pinning control, provider exclusion, bring-your-own-key or own-cloud deployment is documented. 5 1
The Data Protection Officer
The captured pages never say which models sit behind the AI Agent or who runs them, and I found no EU-only processing option an administrator can set across models, agents and connectors. The single location signal is region-specific webhook endpoints for AI Agents in US, EU and Canadian regions in one connector's documentation, which says nothing about where inference happens. 5
The Automation Lead
The captured pages do not name the models behind the product or who runs inference; the only processing statement is that model data is held in memory for the duration of a request under zero-data-retention principles. Integration documentation distinguishes US, EU and Canadian AI Agent endpoints, which hints at regional instances, but I found no public information on inference region, an admin setting to pin processing to the EU, or provider choice. 1 5
The Customer-Service Director
The captured pages show AI agents served from United States, European and Canadian regions, but they name no model and no provider, and say nothing about where inference runs or whether processing can be pinned to the EU. For forty agents' worth of customer conversations, I need to know who processes the prompts, and the pages do not tell me. 5 1
The Works Council Member
The documentation shows separate US, EU and CA (Maple) instances for AI agents, but I found no public information on which models sit behind the agent, who runs them, or where inference happens — zero-data-retention processing is described without naming any provider. Nothing documents pinning processing to the EU, excluding a provider, or bringing our own model. 5 1
The Skeptic
The captured pages name not one model or provider behind this product: no inference region, no admin option to pin processing to the EU, no way to exclude a provider, and no bring-your-own-key or own-cloud deployment. The zero-data-retention wording confirms prompts reach some model whose identity and location the vendor leaves unstated; the only regional signal is per-region webhook endpoints for US, EU and Canadian environments, which says nothing about where answers are generated. 1 5
Admin, identity & access controls
Show reasoningHide 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.
The Head of IT
SAML 2.0 SSO with attribute mapping and session limits is documented, and the audit log is genuinely deployable — every configuration change carries actor, activity, entity and interface (dashboard, API, MCP, import, export) with filters and CSV export capped at 10,000 rows. But provisioning is documented as manual: admins delete existing accounts and re-add them with SSO as the login type, initial SSO setup goes through an Ada representative, and I found no public information on SCIM, on groups governing which models, agents or connectors each team may use, or on streaming logs to a SIEM. 3 4 7
The Data Protection Officer
SAML 2.0 single sign-on with any identity provider, session limits for security compliance, and a genuine audit log recording actor, entity, activity and interface across configuration objects from knowledge sources to API tools — filterable and exportable as CSV, capped at 10,000 rows. But provisioning is a manual delete-and-re-add of user accounts, I found no SCIM, no group-based control over which agents, actions or connectors each team may use, and no documented scope for usage reporting on staff. 3 4 7
The Automation Lead
SAML 2.0 single sign-on is documented (Okta acting as service provider), with a role hierarchy implied by an "Agent or higher" edit-permission rule and session limits for compliance. The audit log is the strongest piece — a chronological, filterable record of who changed what across playbooks, knowledge sources, API tools, processes and API keys, marking whether the change came from the dashboard, API or MCP, exportable as CSV. Provisioning is manual (accounts deleted and re-added with SSO as login type), and I found no public information on SCIM, group-based permissions or documented usage analytics. 3 4 7
The Customer-Service Director
We get SAML single sign-on, a role hierarchy with edit permissions, session limits, and a rich audit log covering knowledge sources, API tools and handoffs, filterable by actor and exportable to CSV. Provisioning is manual — accounts are deleted and re-added to move them onto SSO — and I found no documented group-level permissions for connectors or scoped usage analytics, so it falls short of the governance a regulated contact centre runs on. 3 4 7
The Works Council Member
SAML 2.0 single sign-on with attribute mapping, role-based edit permissions, session limits and a searchable audit log — filterable by actor, activity, entity and interface, exportable as CSV — give administrators real oversight of configuration changes. Provisioning is manual, with accounts deleted and re-created to move them to SSO, and I found no public information on usage analytics or reporting that could be aggregated or pseudonymised for a works-council agreement. 3 4 7
The Skeptic
SAML 2.0 single sign-on with any identity provider, an audit log recording who changed what, when and through which interface with filters and CSV export, session limits, and role-gated setup permissions are real IT-grade material. But provisioning is manual — accounts are deleted and re-added to move them to single sign-on, and initial setup itself needs an Ada representative — and we found no public information on automated SCIM provisioning, group-level permissions over models or connectors, or data-loss-prevention policies; audit exports also cap at 10,000 rows. 3 4 7
Integrations, connectors & knowledge
Show reasoningHide 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.
The Head of IT
The connector set is deep for support knowledge: Zendesk, Salesforce, ServiceNow, Dynamics, Contentful, Dixa, Gladly and Helpjuice, with incremental sync (15 minutes for several sources), private-article ingestion, multiple connections per source, webhooks, a Knowledge API for unsupported sources, handoffs to Zendesk and Salesforce live agents, a voice channel, and MCP listed as a change interface in the audit log. I found no public information on file-store connectors such as SharePoint, or on end-user-facing source citations for grounded answers — which is where a company-wide assistant has to go next. 5 6 7 3
The Data Protection Officer
The connector library covers the support stack — Zendesk, Salesforce, ServiceNow, Dynamics, Contentful, Dixa, Gladly, Helpjuice — with stated automatic sync intervals down to every 15 minutes, article-level availability rules, a Knowledge API for custom sources, webhooks, and handoffs into Zendesk and Salesforce live chat; MCP even appears among the audit-log interfaces. I found no public information on mail or file-store connectors, and while the reasoning log shows retrieved snippets to administrators, I found nothing on citations shown to end users on grounded answers. 5 3 7
The Automation Lead
Eight knowledge-source connectors are documented (Zendesk, Salesforce, ServiceNow, Microsoft Dynamics, Contentful, Dixa, Gladly, Helpjuice) with explicit sync intervals as fast as 15 minutes, private-article and channel-visibility selection, multiple Zendesk and Salesforce connections, and a Knowledge API plus webhooks for anything uncovered. MCP appears as a tracked interface in the audit log, and handoffs to Zendesk or Salesforce chat (or a custom Solutions-team build) close the loop to live agents. Coverage skews to help centres rather than mail or generic file stores, and source attribution rests on a reasoning log showing the search terms and snippets retrieved. 2 3 5 6 7
The Customer-Service Director
Zendesk, Salesforce, ServiceNow, Dynamics and several knowledge bases sync automatically — some every fifteen minutes — with private-article support, availability rules, multiple help-centre connections and a Knowledge API for unsupported tools. Handoffs to Zendesk and Salesforce chat and messaging are documented, plus custom setups through their Solutions team; I found no file-store or mail connectors and no statement that grounded answers show their sources to the customer. 5 6 7
The Works Council Member
Eight knowledge-base and service-desk connectors are documented with concrete sync cadences — fifteen minutes for Salesforce, ServiceNow and Dynamics, two to six hours for others — plus private-article ingestion, availability rules, a Knowledge API for custom sources, and per-conversation visibility of the exact search terms and retrieved article snippets. What is documented here covers knowledge bases, CRM and ticketing; I found no public information on file-store or mail connectors, and nothing describes citations shown to the end user. 5 6 2
The Skeptic
Eight knowledge-base connectors with stated sync cadences — Salesforce, ServiceNow and Microsoft Dynamics every fifteen minutes, Helpjuice every two hours — plus private-article ingestion, availability rules, a Knowledge API for custom sources, webhooks, and handoffs into Zendesk and Salesforce chat is a serious connector story, and MCP appears as an interface in the audit log. Coverage is helpdesk knowledge, though: we found no public information on file-store, mail or CRM-record connectors, and no statement that end users see source citations — the log of retrieved article snippets is operator-facing. 2 5 6 7
Agent capabilities & guardrails
Show reasoningHide 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.
The Head of IT
Playbooks compose conditions, actions and responses into multi-step workflows without code, Actions call external APIs behind tokens with testable sample values, each run logs the request and response data, and there is a test mode plus Test Cases and Simulations tracked in the audit log, with moderation filters and handoffs to human agents. I found no public information on human approval gates before consequential actions, on triggers and schedules, or on vendor-stated limits and failure handling. 7 8 2 3
The Data Protection Officer
Actions call external systems through stored tokens, Playbooks chain conditions, actions and responses into multi-step workflows built without code, and every run is logged with the request sent, the response received and the knowledge snippets retrieved; Actions can be tested with sample values before use. I found no public information on human approval before consequential actions, on triggers or schedules, or on stated limits and failure handling — without those I cannot call this production-grade. 7 8 2
The Automation Lead
Actions call external APIs with token-based authentication, can be tested with sample values before use, and Playbooks compose conditions, actions and responses into multi-step workflows without code — each run showing request data, response data and success status in the conversation view. Testing exists via a Test conversation mode plus Test Cases and Simulations tracked in the audit log, with handoff to live agents supported. I found no public information on a human-approval gate before a consequential action fires, or on triggers and schedules beyond a passing mention of proactives — the guardrail I would want before letting this write into an ERP. 2 3 7 8
The Customer-Service Director
Actions let the agent call external systems with secure tokens, Playbooks chain conditions, actions and responses into multi-step workflows without code, and every action run is logged with the request and response data — the accountability I need around order changes. Testing is a sandbox chat and handoffs to live agents are listed, but I found no configurable human approval before consequential actions, no evaluation against test sets before release, and no vendor statement of what the agent must not be trusted with. 7 8 2
The Works Council Member
Actions call external APIs under token authentication, Playbooks run multi-step workflows, every run logs the request and response data and whether it succeeded, and a conversation can be handed over to a live human agent on Zendesk or Salesforce. Test conversations and sample-value action testing exist, but I found no public information on a configurable human approval gate before consequential actions — my precondition for any agent that touches people — nor on triggers, schedules, or a stated limit on what the agent must not be trusted with. 7 8 2 3
The Skeptic
Actions call external APIs with token-based authentication, Playbooks chain conditions, actions and responses into multi-step workflows built by dragging blocks rather than writing code, every action run logs the request and response data, and test conversations, test cases and simulations appear as tracked objects. We found no public information on configurable human approval before consequential actions, on triggers or schedules, or on the vendor's own statement of what these agents must not be trusted with. Given the agent can be wired into external systems, those are exactly the questions I would ask before production. 2 3 7 8
European sovereignty
panel opinion
Show reasoningHide 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.
The Head of IT
An EU instance region is visible in the connector endpoints alongside US and CA, and a Trust Center retention policy is referenced. But the captured pages give no contracting jurisdiction, no hosting commitment, no subprocessor list including model providers, and no data processing agreement — the sovereignty fundamentals are all listed as unknown. 5 1
The Data Protection Officer
The captured pages give no contracting entity or parent company, no hosting infrastructure, and no subprocessor list — for an AI product the model providers are the first subprocessors I need named with their regions, and I found no public information naming them. The only signal is region-specific webhook endpoints for AI Agents in US, EU and Canadian regions, which does not amount to a hosting or residency commitment. 5
The Automation Lead
I found no public information on the contracting entity's jurisdiction, where customer data is hosted, or the subprocessors — above all the model providers, which the captured pages do not name. The only regional signal is documentation separating US, EU and Canadian "Maple" AI Agent endpoints; a trust centre with a retention policy exists but states nothing about location or safeguards. 1 5
The Customer-Service Director
The only fact I can stand on is region-specific endpoints showing agents in the United States, the European Union and Canada; the contracting entity, the subprocessors and above all the model providers behind the product are undocumented on these pages. I cannot assess from this whether European customers' conversations stay in a European supply chain. 5
The Works Council Member
The one sovereignty-relevant fact I can rely on is that AI agents run in separate US, EU and CA (Maple) instances. Beyond that, the contracting entity's jurisdiction, hosting infrastructure, subprocessors and the model providers behind the product are unstated everywhere I read, and I found no public information on a data processing agreement. 5
The Skeptic
Per-region webhook endpoints for US, EU and Canadian environments hint at regional separation, and the vendor states model processing follows zero-data-retention principles in memory. Beyond that, the captured pages state no contracting-entity jurisdiction, no hosting location for customer data, and no subprocessors — the model providers behind the product go unnamed — and we found no public information on a data processing agreement. Unproven is not sovereign. 1 5
Pricing transparency
not rated — the vendor publishes no price
Show reasoningHide 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.
The Head of IT
The only cost signals on any captured page are that handoffs "may include" options "[d]epending on your plan and tools in use", that "Voice and Social channels may need to be enabled separately for billing reasons", and that initial SSO setup requires coordination with an Ada representative. I found no public information on tier prices, per-seat or per-resolution rates, usage limits, minimums or term. 7 4
The Data Protection Officer
No price figure appears on any captured page — nothing per seat, per conversation or per resolution, no tiers, no limits, no term. The pricing-adjacent statements that do exist — voice and social channels may need separate enabling for billing reasons, handoffs depend on your plan, and initial SSO setup requires an Ada representative — describe a sales-led motion a company cannot budget from public pages. 7 4
The Automation Lead not rated — no public information
I read the captured documentation pages and found no public information on prices, tiers, seat minimums, usage limits or term. The only billing-related statement notes that Voice and Social channels may need to be enabled separately "for billing reasons", which tells a buyer nothing about what anything costs. 7
The Customer-Service Director
The only pricing signals are that Voice and Social channels are enabled separately for billing reasons and that handoffs depend on your plan; I found no public information on any price per conversation, per resolution or per seat, and single sign-on setup itself requires an Ada representative. My CFO cannot forecast a single month from these pages. 7 4
The Works Council Member
The only pricing information is structural: "Voice and Social channels may need to be enabled separately for billing reasons", and handoff options depend on your plan. No price figures, seat or usage units, tier inclusions, minimums or term appear on the pages I read, and even single sign-on setup requires coordination with an Ada representative. 7 5
The Skeptic
No price appears on any captured page — no per-seat figure, no per-resolution figure, no tier table. We learn only that plans exist, that Voice and Social channels may need to be enabled separately for billing reasons, that handoffs depend on the plan, and that single sign-on setup and custom integrations run through an Ada representative. I cannot compute a thousand seats from this, and neither can a buyer. 4 7
European sovereignty — proven facts
0 of 6 dimensions provenBuilt only from facts shown on the vendor's own pages. A dimension we could not prove is left open, not scored as zero.
| Legal entity | Not determined | — | uncited Report an error |
|---|---|---|---|
| Hosting location | Not determined | — | uncited Report an error |
| Hosting options | Not determined | — | uncited Report an error |
| Open source | Not determined | — | uncited Report an error |
| Data processing agreement | Not determined | — | uncited Report an error |
| Transfer impact assessment | Not determined | — | uncited Report an error |
Questions about Ada
Is Ada European?
Not determined: the vendor’s pages as captured do not say where Ada is based.
Where does Ada store data?
Not stated on the vendor’s pages as of 3 Nov 2026.
Does Ada have a free plan?
Not stated on the vendor’s pages as of 3 Nov 2026.
What are the best alternatives to Ada?
As of 4 Nov 2026, the panel rates Langdock (4/5), Dust (3.5/5) and Unique AI (3.5/5) highest among the other 18 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). Ratings are the AI panel’s opinion, formed from public evidence: how scores are made. The full list, with European products first: Ada alternatives.
Where this could be wrong
- Evidence ages. The oldest capture behind this page is from 3 Nov 2026. Vendors change pricing and policies without notice; every fact reflects its source as of the capture date shown in the registry.
- AI can misread a source. Extraction and judgement are automated; a citation guarantees traceability, not infallibility. If something here is wrong, say so — no account needed, every report is decided within 5 business days, and accepted corrections are published.
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.
- We found no public information on pricing on the pages we read (docs.ada.cx/docs/other/data-retention, docs.ada.cx/docs/optimization/conversations/response-training, docs.ada.cx/docs/other/audit-log, docs.ada.cx/docs/other/team-access/sso, docs.ada.cx/docs/knowledge/content-ingestion/knowledge-integration, docs.ada.cx/2026-03-18-salesforce-knowledge-v2 and 2 more). If the vendor publishes it somewhere else, send us the page. Know more? Tell us
- 1 product fact could not be confirmed on the vendor’s page as captured and was left out of this page and of the panel’s material. Know more? Tell us
Sources (8)
The pages every claim on this page was read from — each one checked, dated, and kept verifiable.
- 1 Data use, training & retention — found from sitemap docs.ada.cx Checked 3 Nov 2026 Details →
- 2 Data use, training & retention — found from sitemap docs.ada.cx Checked 3 Nov 2026 Details →
- 3 Admin, identity & access controls — found from sitemap docs.ada.cx Checked 3 Nov 2026 Details →
- 4 Admin, identity & access controls — found from sitemap docs.ada.cx Checked 3 Nov 2026 Details →
- 5 Integrations, connectors & knowledge — found from sitemap docs.ada.cx Checked 3 Nov 2026 Details →
- 6 Integrations, connectors & knowledge — found from sitemap docs.ada.cx Checked 3 Nov 2026 Details →
- 7 Agent capabilities & guardrails — found from sitemap docs.ada.cx Checked 3 Nov 2026 Details →
- 8 Agent capabilities & guardrails — found from sitemap docs.ada.cx Checked 3 Nov 2026 Details →