Nothing we track differs between them.
| Capability | Bland AI | Lorikeet |
|---|---|---|
| Not published | IncludedAll plans. Billed at 0.95 credits per resolution on Start, 0.80 on Scale. | |
| Live chat | IncludedEnterprise only per the compare table, though the billing docs publish the rate anyway at $0.01 per agent message. The widget carries both voice and chat and uses the same pathways, personas and memory as the phone channel. | IncludedAll plans. Lorikeet Chat deploys the agent on a website without a ticketing system behind it. |
| Voice | IncludedStart. Voice is the product, billed per connected minute. | IncludedAll plans, at 1.50 credits per resolution on Start and 1.20 on Scale. |
| SMS | IncludedEnterprise only, priced at $0.02 per message in the billing docs. US numbers need A2P registration, and BYOT numbers must have their A2P campaign set up in Twilio because Bland cannot yet do it. | IncludedAll plans, priced identically to chat and email. |
| Capability | Bland AI | Lorikeet |
|---|---|---|
| Autonomous resolution agent | Not published | IncludedThe Concierge Agent is the product. Billing exists only for resolutions and per-ticket routing or QA actions. |
| Resolution over knowledge base | IncludedCounted rather than metered: 10 knowledge bases on Start, 50 on Build, 100 on Scale, unlimited on Enterprise, with no per-query charge. Built from an uploaded file, raw text or a sitemap crawl. | IncludedKnowledge is read in place from Confluence, Guru, Google Docs and Notion. No knowledge base migration is required. |
| Action-taking (API calls) | IncludedTools call external APIs mid-conversation, webhook nodes fetch data inside a pathway, and secrets hold credentials without hardcoding them. Expanded call logs show every webhook URL, status code and response payload after the fact. The Custom Code node, which runs JavaScript inside a pathway, is Enterprise. | IncludedConcierge actions and data integrations are listed as unlimited on every plan including Start. |
| Handoff to human | IncludedWarm Transfers and Live Transfers are both Enterprise on the compare table. Cold transfer to a number is available lower down and bills the transferred leg at the plan's transfer rate, or free on your own Twilio. | Not publishedLorikeet never writes the words escalation or handoff in its pricing or product text. Whether a handoff exists, and whether it bills, are both unpublished. |
| Guardrails & citations | IncludedBoth Enterprise on the compare table. Guard rails run on every AI and user response in production and come in two shapes: time-based rules that must fire (AI disclosure, self-introduction, recording disclosure, default window 30 seconds) and content rules that must not. Citations extract structured values from transcripts with the quoted source text and timestamp attached. | IncludedCustom guardrails are unlimited on every plan including Start; Enterprise adds custom guardrail work. |
| Multilingual AI | Included17 languages named on the current voice generation plus 10 marked experimental. The docs index claims 40+ out of the box, which is not the same number as the enumerated list. Live Translation streams audio and returns translated speech between supported pairs. | Not publishedNo language count or list is published anywhere on the site. |
| Capability | Bland AI | Lorikeet |
|---|---|---|
| Auto-tagging | Not published | IncludedAll plans, billed separately at 0.30 credits per ticket on Start and 0.25 on Scale. |
| Capability | Bland AI | Lorikeet |
|---|---|---|
| Voice agent | IncludedTelephony arrives three ways: numbers bought from Bland at $15 a month, your own Twilio account, or SIP trunks from Genesys, Five9, NICE CXone, Amazon Connect or Talkdesk. SIP traffic bills at the standard voice rate with no SIP surcharge. | IncludedWorks with any platform that supports call forwarding, including Amazon Connect, Twilio, Genesys Cloud, Dialpad, Vonage, Five9 and AirCall, plus the Lorikeet SDK and API. |
| IVR replacement | IncludedA named solution page, and pathways handle DTMF button presses as call events. Bland says the platform was built phones-first, which shows in features like custom dialing that picks the outbound number to maximise pickup. | Not published |
| Real-time transcription | IncludedBland's own speech-to-text rather than Deepgram or a reseller, which is why no transcription vendor appears in the sub-processor list. Live transcripts stream in the Active tab of call logs, and a separate endpoint returns a corrected transcript re-derived from the recording afterwards. | Not published |
| Call summarization | IncludedAn AI summary is generated per call and returned on the call object, so it comes out with the export rather than living only in the dashboard. | Not published |
| Capability | Bland AI | Lorikeet |
|---|---|---|
| Automated QA coverage | IncludedEvals grade real calls at scale with configurable LLM judges, and every call carries automatic metrics for agent quality, transcription quality, sentiment, silence, interruptions and repetition, with issues flagged by severity. A human review workflow sits alongside it: assignee, status, QC tags and threaded notes. No export path is documented for Evals output. | IncludedAll plans, billed at 0.30 credits per ticket on Start and 0.25 on Scale. |
| Capability | Bland AI | Lorikeet |
|---|---|---|
| Raw data export | IncludedRare in this category: an actual export button, not just an API. Export Calls in the call logs header mails a CSV, respects the filters on screen, spans up to 180 days and lets you include or exclude individual fields. | Not publishedNo export documentation of any kind is published: no article, no API reference, no bulk export UI. |
| Capability | Bland AI | Lorikeet |
|---|---|---|
| SAML SSO | IncludedEnterprise only. Both OIDC and SAML 2.0, with a setup wizard covering Okta, Azure AD, Google Workspace and Auth0. Users from a verified domain are provisioned automatically on first login. | Not published |
| HIPAA / BAA | IncludedBAA is Enterprise only, blank on all three published columns. HIPAA eligibility is claimed without a tier, but a regulated buyer cannot sign the agreement that makes it usable at a published price. | IncludedScale and above, published as "Start + HIPAA BAA". Explicitly not available on Start at $1,500/month. |
| SOC 2 | IncludedNo plan gate on the certification. Type I and Type II, with quarterly third-party penetration tests in scope. The report itself is behind the Trust Portal, and the pricing FAQ adds that compliance documentation is available under NDA on Enterprise. | IncludedStart and above. The pricing page specifies SOC 2 Type 2. |
| ISO 27001 | Not publishedNot claimed on the trust page, the AI information page or the pricing FAQ, all three of which enumerate certifications and stop at SOC 2, HIPAA, GDPR, PCI DSS and FedRAMP. | IncludedStart and above. ISO 27001:2022. |
| GDPR | IncludedNo plan gate. A public DPA incorporates the 2021 EU Standard Contractual Clauses and the UK IDTA, and covers CCPA, the Colorado, Connecticut, Virginia and Utah statutes and the Swiss FADP alongside GDPR. Sub-processor changes carry 14 days notice and an objection right. | Not publishedGDPR is not named on the trust center, the privacy policy or the pricing page. |
| Data residency | IncludedEnterprise only for selection, though the regions themselves are documented publicly: Oregon, Canada Central, Sydney and Ireland, each with its AWS region and static egress IPs. Owning the models is what makes the claim different from a wrapper's, since inference runs on hardware in the region rather than routing to a US provider cloud. | Not publishedNo customer-selectable region is published. Infrastructure is Google Cloud Platform and all four listed sub-processors are in the USA, so there is no EU-only inference path. |
| Data retention policy | IncludedNo plan gate in the API reference. recording_lifespan_days is set per organization from 1 to 1825 days, or -1 to disable retention. It governs recordings only; no equivalent control is documented for transcripts, pathway logs or extracted variables. | Not publishedThe trust center states "Data retention procedures established" and "Customer data deleted upon leaving" but publishes no window in days. |
| Right-to-erasure tooling | IncludedUnder the DPA, Bland deletes or returns all Customer Personal Data at the end of the Services on request, excluding backups which it isolates from further processing. The request window is 14 days from cessation, by email to security@bland.ai. | Not published |
| Capability | Bland AI | Lorikeet |
|---|---|---|
| REST API | IncludedNo plan gate. A dashboard API key works from Start. | IncludedThe published API initiates voice conversations, and two public HTTP endpoints handle demo booking and support Q&A. No API reference covering data extraction exists. |
| Webhooks | IncludedPost-call webhooks fire after a call ends and can be fetched, created or resent through the API. Payloads are verifiable by HMAC signature, and Enterprise adds asymmetric JWT signing. The CLI forwards webhooks to localhost without ngrok. | Not published |
| SDKs | Includedbland-cli from npm covers the whole account and emits JSON on every read command. It bundles a local MCP server and Bland hosts a remote one. A Web Agent SDK embeds an agent in a browser. No first-party Node or Python server library is documented. | IncludedThe Lorikeet SDK embeds the agent in a customer's own product. |
| Published rate limits | IncludedPartly. org_rate_limit is documented with a default of 5, and the send-call reference enumerates five 429 classes including a 10-second floor between calls to the same number and separate limits for new accounts and international dialing. There is no per-endpoint table, and none of it covers the read endpoints used to extract data. | Not published |
Both entries last checked Sep 9, 2026.
2 of 9 not published by Bland AI, checked
6 of 9 not published by Lorikeet, checked