Relevance AI: building AI agents without your own development team

AI agents have arrived in mid-sized companies, but implementation still stalls. Anyone who wants to go beyond simple if-then workflows soon faces a question: how do you build AI-driven processes that make decisions on their own, without first assembling a development team? Relevance AI promises exactly that. The Australian platform positions itself as an “AI Workforce Platform” on which business users can create and orchestrate autonomous AI agents with low-code tools.

Does it deliver on that promise? We look at what Relevance AI actually does, where its strengths lie, and where you should keep your expectations realistic.

Relevance AI – as of April 2026

Status:

• New enterprise analytics introduced, with detailed usage analysis per team member.
• Documentation completely overhauled; new LLM options (including Gemini 3 Flash Preview) available.
• Series B funding closed with well-known investors (Bessemer Venture Partners, Insight Partners, AWS); the company is on solid financial footing.
• Users often report unpredictable credit consumption and a steep learning curve for more complex agents.

What is Relevance AI?

Relevance AI is a low-code platform for building and orchestrating AI agents. The company was founded in Sydney in 2020 and mainly targets go-to-market teams in sales, marketing, and operations.

The key difference from classic automation tools such as Zapier or Make: in Relevance AI, agents are given a goal and decide for themselves the order of steps, which tools to use, and how to process information. Instead of rigid if-then rules, the platform relies on embedded reasoning.

The platform does not build its own language model; it is LLM-agnostic. You choose from models by OpenAI (ChatGPT), Anthropic (Claude), Google (Gemini), or Cohere, and you can bring your own API keys. Through a RAG (retrieval-augmented generation) system, agents draw on company-specific knowledge bases, which reduces hallucinations and enables domain-specific answers.

What is Relevance AI good for?

Automated sales development (AI BDR)

For companies with small sales teams, this is the strongest use case. AI agents research potential leads on the web, gather context such as company size, recent news, or tech stack, and write personalized outreach messages.

Several agents work together as a team in a multi-agent system: a research agent gathers the context, an AI BDR writes the outreach, and an AI SDR qualifies incoming replies.

From practice: The team at Verisoul (two people) reported that with Relevance AI it operated like a 20-person team, generating five to ten new leads and prepared meeting materials per day with less than five minutes of review time daily (source: Relevance AI).

→ A mid-sized company with five to ten sales reps can therefore build significantly more pipeline without having to create new positions right away.

Automated inbound lead qualification and routing

Incoming leads from website forms, trade show contacts, or partner referrals are qualified automatically as soon as they arrive. The agent checks them against defined criteria such as industry, company size, or budget signals, and routes qualified leads to the right sales rep. CRM updates happen automatically.

For companies that generate leads across several channels, this means no missed opportunities because of slow response times. According to the vendor, Relevance AI has used this approach to fully automate its own internal SDR function.

Research and analysis of unstructured data

Agents sift through large volumes of information, from documents and customer feedback to market reports and competitor websites, and extract structured insights. RAG-based knowledge bases let you bring in company-specific data such as PDFs, CSVs, or SharePoint content.

For companies that regularly run competitive analyses, product research, or document reviews, this can cut hours of manual work to minutes. It is especially relevant for product management and strategic planning, wherever unstructured information has to be distilled into a basis for decisions.

AI-assisted first-level support

AI agents answer common customer questions from a knowledge base, categorize tickets, and escalate complex cases to human staff. This works well for structured, FAQ-style requests and relieves the support team of standard questions.

One clear caveat, though: Relevance AI is not the best choice for comprehensive customer support. Dedicated support platforms such as Zendesk AI or Intercom Fin are faster to deploy and more deeply integrated for complex support flows. Here, Relevance AI works as a complement, not as the primary solution.

Our assessment

Where it pays off

Relevance AI is one of the most capable platforms for agent-based automation. Its combination of LLM agnosticism, multi-agent orchestration, and low-code access is rare. For comparison: Zapier and Make cannot reason, while CrewAI and n8n are too technical for pure business users.

Companies with small to mid-sized sales teams that want to scale outreach without adding headcount benefit most.

If you already use Salesforce, HubSpot, or Microsoft 365, you will find native integrations that work.

And if compliance requires you to be able to switch between LLM providers, this is one of the few platforms that takes that seriously.

Where it is the wrong choice

Despite the “no-code” positioning, community feedback paints a clear picture: anything beyond simple workflows requires API knowledge, prompt engineering, and iterative debugging.

Cost predictability is a real issue. Always run a pilot with real data before committing larger budgets.

Debugging has its limits too. Agents do not produce a detailed thinking log. When an agent makes a wrong decision or gets stuck in a loop, troubleshooting comes down to trial and error rather than root-cause analysis.

The platform is not suited to companies that expect immediate results. Realistically, production-ready agents take several weeks to a few months to implement.

Lucid Labs recommendation

We recommend Relevance AI for mid-sized companies with ambitious sales targets and at least one technically skilled person on the team. Do not take the platform’s “no-code” promise literally; see it for what it is: low-code with a learning curve.

The AI BDR use case (automated sales development) is the clearest quick win. Start with a single agent for a clearly defined use case, bring your own API keys (this avoids the platform markup), and budget a buffer for unexpected credit consumption.

Current prices are listed on Relevance AI’s official pricing page.

FAQ

Do you need programming skills to use Relevance AI?

Not necessarily for simple agents and prebuilt templates; the low-code builder can be used without coding. For production-ready multi-agent workflows, custom API integrations, and debugging agents that misbehave, however, a basic technical understanding (API knowledge, prompt engineering) is practically indispensable. The learning curve is steeper than the marketing suggests.

How secure is the data, and can you use EU data centers?

Relevance AI is SOC 2 Type II certified and GDPR-ready. You can choose between US, EU, and AU data centers, so data can stay within the EU. For industry-specific requirements (in healthcare, for example), check before deployment whether the available certifications are sufficient for your sector.

Which language models can you use?

The platform is LLM-agnostic. Supported models include those from OpenAI, Anthropic, Google, and Cohere. You can bring your own API keys (recommended to reduce costs) or use the platform’s keys, in which case a markup on model costs applies. You can also use different models for different tasks within the same workflow.

How long does it realistically take for an agent to be production-ready?

For a simple, clearly defined use case (such as lead research with a template), working results are possible within days. For production-ready multi-agent workflows with custom integrations, community members report weeks to months of iterative refinement.