Odella

Odella's defining product model is “employment-shaped” automation. A team can hire a specialist from the marketplace or configure one in plain language, connect only the tools that worker needs, assign responsibilities and schedule work. Workers can respond to events, run long jobs in the background and operate during contracted working hours without a user prompting every step.

The product advertises more than 100 integrations on the current homepage and a broader 1,000+ managed-app connection surface on the features page. Named examples include Gmail, Slack, Google Drive, Salesforce, HubSpot, Xero, Notion and Stripe. Workers can also be reachable through Slack, Microsoft Teams, email, WhatsApp, Telegram, phone and SMS. Dedicated phone numbers and document-filling/signature workflows extend the product beyond browser-only automation.

Governance is unusually prominent. Each worker runs in an isolated environment; access can be scoped to a particular inbox, folder or account; owners/admins/members have differentiated permissions; outbound or sensitive actions can require human approval; and work history is logged for review. Odella describes enterprise practices aligned to SOC 2 and ISO 27001 and separately uses the wording “SOC 2-grade practices.” The public wording should not be interpreted the public site as proof of a completed independent SOC 2 or ISO certification unless the underlying assurance artifact is produced.

Odella also supports structured, no-code workflows for work that should be predictable rather than fully autonomous. Its product materials describe triggers, tasks, conditions and automation blocks, while document tooling supports DOCX/PDF templates and knowledge stores. This hybrid of structured workflow and open-ended agent execution is a sensible architecture for regulated or repeatable knowledge work, although independent reliability testing was not found.

Odella uses hourly, pay-as-you-go pricing rather than per-seat subscriptions. Customers set contracted working hours for each AI employee and choose worker capacity/model; charges accrue for scheduled hours while the worker is enabled. The official calculator changes dynamically with configuration, so there is no single universal monthly price. The site offers free onboarding/start access with no credit card, no human-seat fee, and the ability to pause or cancel. Current third-party launch listings cite a starting rate around $0.40 per worker-hour, but buyers should use the official calculator because actual hourly cost depends on model and machine configuration.

The reviewed source set produced no substantive exact-product editorial review, benchmark, mature software-review corpus or eligible numeric rating for Odella. Five reviewed public source domains were technically inaccessible. Apple App Store and Google Play exact-name searches returned apps associated with a developer named Oriol Della rather than Odella AI; those results were excluded for product-surface mismatch. Other reviewed-source searches returned no relevant exact-product evidence. Missing coverage is not scored as zero, but it reduces confidence.

Outside the main reviewed source set, current third-party evidence is mostly ecosystem and launch corroboration rather than testing. LinkedIn company data identifies Odella as a Sydney software company founded in 2025, and launch posts in August 2026 confirm the current AI-employee positioning. Smokeball's marketplace lists Odella as an integration for legal practices and describes document/process automation against Smokeball data. These sources corroborate identity and use cases but do not independently benchmark output quality.

Privacy documentation is materially more detailed than many early agent platforms. LAIW PTY LTD is identified as the operator. The Privacy Policy says customer data in active systems is deleted within 30 days after account deletion or a valid request, backups ordinarily expire within 90 days, and security/access/operational logs are generally retained up to 24 months. A July 2026 DPA describes Odella as Processor/Subprocessor for customer personal data and incorporates appropriate contractual transfer mechanisms where applicable.

Odella's LLM-training policy deserves precise treatment. Customer Data, Content and AI Agent Traces are not used to train or fine-tune Odella-developed LLMs unless an authorized organization owner expressly opts in. However, the policy separately permits service-improvement processing such as model routing, workflow reliability, safety evaluation and product improvement, and third-party model/API providers can have their own data terms. Organizations with sensitive data should therefore review configured providers and subprocessor terms rather than treating “no training by default” as a blanket no-processing statement.

The security surface includes least-privilege access, isolated worker environments, approval gates, SSO/RBAC-style administration, regional residency options, PII removal claims and audit logs. These are meaningful controls. The remaining evidence gap is assurance: The reviewed public evidence does not include a SOC 2 report, ISO certificate, penetration test or external reliability benchmark during this desk review.

Odella is strongest for professional-services, legal, compliance, finance, operations, support and other knowledge-work teams that have repeatable processes spread across email, documents, CRM and collaboration tools. It is especially relevant where organizations want a non-technical employee-like interface but still require scoped permissions and human approval for consequential actions.

It is less compelling for teams that only need occasional chat assistance or deterministic app-to-app automation: a general assistant or conventional workflow engine may be simpler and cheaper. Buyers that need independently validated reliability, a large public user-review corpus or completed public assurance documents should treat Odella as an early-stage platform requiring a structured pilot and procurement review.