Comparison
Syniaps or Relevance AI?
A six-year-old Australian scale-up, a no-code agent workshop sold to sales teams, now turning towards assisted selling
What Relevance AI sells: An "AI Workforce": agents built and customised by your team, routing to the cheapest model that passes your tests, 2,000 integrations, SOC 2, and a promise of tasks at $0.09 each.
Syniaps
4.0 / 5
Relevance AI
3.0 / 5
Surveyed on 15 September 2026 from the publisher's public pages, cited at the foot of this page. Offers change: check with the publisher before you decide.
Relevance AI at a glance
- Who
- Sydney, 2020, $37m raised
- Relevance AI Pty Ltd, founded in 2020 in Sydney with a second office in San Francisco, has raised $37m in total including a $24m Series B led by Bessemer in May 2025 (Insight, King River, Peak XV). Around 119 employees in mid-2026, 40,000 agents created on the platform by January 2025, SOC 2 Type II and GDPR on display, and named customers including Qualified, Canva, Autodesk, Rakuten and SafetyCulture.
- What
- Agents, tools, workforces, evals
- An agent is a prompt, some tools (LLM steps, Python, API calls, 1,000 to 2,000 integrations), triggers (email, Slack, cron schedule) and a "Knowledge" base (retrieval synced from Google Drive, SharePoint and Notion). A "Workforce" links several agents on a visual canvas with approvals and escalations; "Invent" generates a first agent from a description, and "Evals" tests it through simulated conversations.
- How it gets set up
- You build, then you tune
- You create an account, choose a data region once and for all, then for each agent you write the prompt, pick the tools, wire the triggers, create the knowledge sets and set each link to "Auto run", "Approval required" or "Let agent decide". Invent produces a draft, but the website itself sells a six-week guided programme (mapping the use cases, deploying with the Relevance team, then training the internal teams to build on their own).
- Which models
- Multi-model, routed to the cheapest
- Claude (Haiku 4.5, Sonnet, Opus), GPT 5.4 Mini and 5.5, Gemini 3 Flash and 3.1 Pro, Kimi K2.6, GLM 5.2, plus open models hosted by Relevance on AWS and Azure; the router keeps "the cheapest model that clears your evals". Inference goes to OpenAI and Anthropic through their APIs (with no retention), unless you connect your own AWS Bedrock or Google Vertex account, in which case it runs in the cloud region you chose and outside Relevance credits.
- Price
- $19 to $349 a month, then by quote
- Pro at $19 a month (annual) or $29 (monthly) for 2,500 Actions, $20 of Vendor Credits and 2 builders; Team at $234 (annual) or $349 (monthly) for 7,000 Actions, $70 of credits, 5 builders and 45 users; Enterprise by quote. Two meters: the Action (each tool a agent runs) at $80 per 1,000 on overage, and Vendor Credits (LLM cost at wholesale, $20 per 10,000); the Free plan has been withdrawn from new accounts and the public pricing page now shows only Enterprise with "Talk to sales".
- Who it is for
- Go-to-market teams with builders
- Sales, marketing, customer success and RevOps teams with 2 to 5 people ready to build automations (CRM enrichment, call preparation, outbound prospecting). Strong at scale-ups equipped with HubSpot or Salesforce, much less so at a smaller company with nobody to assemble and maintain the agents.
Point by point
| Criterion | Relevance AI | Syniaps | Advantage |
|---|---|---|---|
| Intelligence | Claude, GPT, Gemini, Kimi, GLM; routed to the cheapest one that passes the evals | Claude Sonnet, Opus, Fable, GPT through Codex or the model of your choice; Standard, In-depth and Maximum modes | Level |
| "Execute" | API tools, Python and integrations; a v0 browser through Airtop, no terminal and no files | Files, code, terminal, a browser with session cookies, long-running work, hosted apps | Syniaps |
| Knowledge | "Knowledge" retrieval: sets to create and attach, synced from Drive, SharePoint and Notion | Synapses built and maintained on their own from documents, email, a cloud mirror and OCR | Syniaps |
| Setup | A prompt, tools, triggers and links to configure; a six-week programme on offer | A short wizard, then the agent proposes the automations in conversation and the human confirms | Syniaps |
| Learning | Memory means metadata fields defined by hand, with no retention of corrections | Retains corrections, skills and preferences, and proposes what it has learnt for approval | Syniaps |
| Sovereignty | SOC 2 Type II, GDPR, a choice of region (including London), multi-tenant, bring-your-own Bedrock or Vertex | A dedicated OVH server in France, a sandbox per user, AI processed in Europe if you choose, no certification | Level |
| Entry and price | $19 to $349 a month then by quote, Actions billed even on failure, Free withdrawn, assisted selling | A free 7-day trial with no human approval, a card required, then from €99 excl. tax a month, server and AI included | Level |
| Scope | Framed go-to-market agents, multi-agent workforces, evals and A/B tests | End-to-end missions run by eleven expert teams, from the document to the deployed website | Syniaps |
What each one does better
Where Syniaps wins
Nothing to build
At Relevance every agent is drawn out (prompt, tools, links, approvals); Syniaps proposes the automations itself after a five-screen wizard.
Full execution
Files, code, terminal, an authenticated browser, background work and app hosting, where Relevance chains API tools and a v0 browser limited to extracting text.
A living company memory
Synapses build and maintain themselves from documents, email and cloud storage, with no knowledge sets to create or attach.
Learns from corrections
Syniaps retains preferences and skills and proposes what it has learnt; Relevance's "Memory" extracts predefined metadata, it does not learn.
A dedicated machine, a readable cost
One server per customer in France and a monthly plan, against a multi-tenant platform billed by the Action (failures included) whose overage surprises teams.
Where Relevance AI wins
Certifications and regions
SOC 2 Type II, GDPR, SAML SSO, RBAC, audit logs, PII masking and a choice of region (US, London, Sydney) from the contract onwards, which Syniaps does not yet have.
Inference in your own cloud
Connecting your own AWS Bedrock or Google Vertex runs Claude in the cloud region you chose, outside Relevance credits and without going through Anthropic's public API.
Routing and evals built in
Evaluations through simulated conversations, a measured success rate, A/B tests and automatic switching to the cheapest model that holds the bar.
Visual multi-agent work
The Workforce canvas links specialised agents with per-link approvals and escalations to a human or another agent, readable by a whole team.
Maturity and network
Six years, $37m raised, 119 employees, 40,000 agents created, a directory of expert partners and references such as Canva, Autodesk and Rakuten.
Two caveats on its promise
- "Zero code" does not mean zero work: the website itself sells six weeks of guidance, memory is configured field by field, and reviews note that "anything with complex branching logic quickly becomes harder to manage".
- The real cost is unstable: two meters (Actions and Vendor Credits), failed runs are billed, overage is $80 per 1,000 Actions, and both the Free plan and self-serve are retreating in favour of "Talk to sales".
The ten criteria, scored and justified
Intelligence
Frontier models (Claude, GPT, Gemini) and ways to step up a gear
Syniaps 5/5
Claude Sonnet 5, Opus 5 and Fable 5, GPT through Codex, or the model of your choice; Standard, In-depth and Maximum modes.
Relevance AI 4/5
Claude Haiku 4.5, Sonnet and Opus, GPT 5.4 Mini and 5.5, Gemini 3 Flash and 3.1 Pro, Kimi K2.6 and GLM 5.2 with automatic routing to the cheapest one that passes the evals; no user-driven step-up mode.
Execution power
Actually acts: files, code, browser, sites without an API, long-running work
Syniaps 5/5
Files, code, terminal, a browser for sites without an API, long jobs in the background, hosted apps.
Relevance AI 2/5
API tools, Python and 1,000 integrations, but the browser is a v0 through Airtop limited to navigation, simple forms and text extraction; no terminal, no files, nothing for sites without an API using the user's session, and no long-running work.
Company knowledge
A base built and maintained on its own from documents, email and cloud storage
Syniaps 4/5
Synapses built by the agent, OCR, a full cloud mirror; still young on very large volumes.
Relevance AI 2/5
The "Knowledge" retrieval store syncs Google Drive, SharePoint and Notion, but each knowledge set has to be created then attached to an agent, with no company base built and maintained on its own.
Zero setup
Nothing to brief, trigger or tick; the tool proposes, the human confirms
Syniaps 4/5
A short wizard, and the agent proposes its automations in conversation; site connections still have to be made once.
Relevance AI 2/5
Invent generates a draft agent from a sentence, but the prompt, tools, triggers, links and approval modes are still yours to set, and the website sells a six-week guided programme.
Continuous learning
Remembers corrections and preferences without being asked to
Syniaps 4/5
Corrections, skills and preferences are retained and proposed for approval; no learning across customers yet.
Relevance AI 2/5
"Memory" extracts metadata according to fields and rules defined by hand ("Let Agent Decide" or "Rule Based"); there is no automatic retention of corrections or preferences.
Autonomy
Missions and automations server-side, computer switched off, human approval on sensitive moves
Syniaps 4/5
Missions and scheduled actions run server-side with your computer off, with approval cards; mission ceilings still need tuning.
Relevance AI 4/5
Cron triggers scheduled server-side, multi-agent workforces, per-link approvals ("Approval required") and escalations to a human through Slack or email, but bounded by the tools that were configured.
Integrations and connections
Everyday tools, MCP, cloud storage, email, sites without an API
Syniaps 4/5
Email, cloud storage, meetings, WhatsApp, LinkedIn, MCP, sites without an API; fewer native connectors than a Zapier.
Relevance AI 4/5
1,000 to 2,000 integrations (HubSpot, Salesforce, Slack, Gmail, LinkedIn, Notion), an API step and Python, bring-your-own Bedrock and Vertex; MCP connectors are undocumented and sites without an API are out of reach.
Sovereignty and security
EU hosting, per-customer isolation, EU or on-premise inference, certifications
Syniaps 3/5
A dedicated machine in France (OVH), strong per-user isolation and AI processed in Europe if you choose; no on-premise option and no certification.
Relevance AI 3/5
SOC 2 Type II, GDPR, AES 256, SSO and RBAC, with a choice of region (US Virginia, "EU" in London and therefore outside the EU, Sydney), but logical separation in a multi-tenant platform, single tenancy "in the works", no on-premise option and inference at OpenAI and Anthropic unless you bring your own cloud.
Teamwork
Shared spaces, roles, delegation, company-wide memory
Syniaps 4/5
Company sources connected once, shared synapses, role profiles, shared skills and conversations, shared mailboxes in missions; no SSO and no fine-grained roles.
Relevance AI 4/5
Team means 5 builders, 45 end users and 5 shared projects, with RBAC and SSO on Enterprise, and agents and knowledge shared inside the project, but no company-wide memory beyond the knowledge sets.
Price and access
Total cost (tool + AI + machine) against what you get, and how easily you start
Syniaps 3/5
A dedicated machine, AI included and a 7-day trial that opens within minutes, from €299 excl. tax a month; a card is required for the trial and there is no free tier.
Relevance AI 3/5
Pro from $19 a month and Team at $234 a month (annual), AI included, but Free withdrawn, self-serve in retreat ("Talk to sales"), $80 per 1,000 Actions on overage, failures billed and unpredictable cost cited as the first complaint.
Verdict
The agent workshop to assemble against the intelligence that already knows the company
Relevance AI
Relevance AI is a mature, certified no-code workshop where a go-to-market team builds and watches its own agents, billed per action.
Natural fit: sales and marketing scale-ups with 2 to 5 builders, a structured CRM and a usage budget.
Syniaps
Syniaps delivers real execution and knowledge of the company with nothing to assemble, on a dedicated machine in France, for a monthly plan.
Natural fit: companies with no builder that want to hand over whole missions rather than draw workflows.
The collision zone is narrow: Relevance wins the teams that want to build, Syniaps the ones that want it done.
Do not chase the multi-agent canvas or model routing: hammer home "nothing to brief, nothing to build" and the depth of execution. Close first the visible certification gap (SOC 2 or ISO, AI processed in Europe already being an option), which is the one Relevance argument an IT director will put to Syniaps.
Public sources (17)
- https://relevanceai.com
- https://relevanceai.com/pricing
- https://relevanceai.com/docs/get-started/pricing
- https://relevanceai.com/docs/admin/security
- https://relevanceai.com/docs/llms.txt
- https://relevanceai.com/docs/integrations/llm-integrations/anthropic
- https://relevanceai.com/docs/build/agents/build-your-agent/memory
- https://relevanceai.com/docs/build/agents/create-an-agent
- https://relevanceai.com/docs/build/workforces/workforce-features/approvals-and-escalations
- https://relevanceai.com/docs/build/agents/build-your-agent/agent-triggers/scheduled-triggers
- https://relevanceai.com/changelog/enable-your-ai-agents-to-browse-and-extract-data-with-browser-automation
- https://techcrunch.com/2025/05/06/relevance-ai-raises-24m-series-b-to-help-anyone-build-teams-of-ai-agents/
- https://tracxn.com/d/companies/relevance-ai/__GcnNqHYzkszu5Cqnv3GUOXxUAPJ0AGlKnzFyOXHCE70
- https://coldiq.com/blog/relevance-ai-pricing
- https://www.quantumdesk.com/blog/relevance-ai-reviews
- https://theaiagentindex.com/agents/relevance-ai
- https://www.capterra.com/p/10019108/Relevance-AI/
Relevance AI is a trademark of its publisher, cited here for the purpose of objective comparison. Syniaps is neither affiliated with nor a partner of that publisher. Found an error in this comparison? Write to us at [email protected] and we will correct it.