AVA

Automation Guide — what the terms mean and where to start

Not sure where automation or AI fits your business?

This guide explains the main types of automation and AI tools in plain English — no jargon, no sales pitch. Use it as a map before you talk to AVA, or come back to it when a term comes up that you don't recognise.

AVA advises — it doesn't build. It helps you understand your options, compare tools, and plan next steps. Once you know roughly what you're looking for, AVA can take it from there.

Ready? Start using AVA

Automation types — in plain language

Most real-world use cases sit somewhere in these six categories. Use them as a map, not a rulebook.

💬

1. Chatbots — informational & conversational

Answer questions like "What time are you open?", "What products do you sell?", or "What are the symptoms of X?" — usually from a fixed knowledge base.

  • Informational: opening hours, pricing, services
  • Conversational: guided Q&A, explanations, symptom info
  • Search-style: "Find places I can stay with a tent"

Typical platforms: Chatbase, Intercom, website widgets, custom HTML embeds.

Ask AVA"I want a chatbot that answers basic questions from my website. What type do I need?"
📥

2. Processors — emails, invoices, requests

Handle inbound items — emails, invoices, forms, refund requests. They read, extract key details, and push data into the right system.

  • Inbox triage — classify and route messages
  • Invoice reading — extract amounts, dates, suppliers
  • Refund/support requests — capture details into your CRM

Typical tools: Power Automate, Zapier, Make, n8n, Pipedream, document AI services.

Ask AVA"I get invoices and emails that I manually copy into different systems. What are my options?"
🔗

3. Orchestrators — multi-step journeys

Link several steps together — show products, take an order, check stock, update systems, trigger delivery.

  • E-commerce flows (show products → order → confirmation)
  • Appointments (check availability → book → reminders)
  • "Do this, then that" chains across multiple apps

Typical tools: Zapier, Make, Power Automate, n8n, Pipedream, Workato, Bubble, Glide.

Ask AVA"Design an end-to-end flow for customers to order via chat and get confirmation emails."
🗂️

4. Workflow management — internal processes

Help your team internally — approvals, task handovers, and multi-step workflows across departments.

  • Approval workflows (leave, expenses, purchasing)
  • Internal task routing (who does what, when)
  • Sales → operations → finance handoffs

Typical tools: Power Automate, Zapier/Make, Jira or service-desk tools, internal apps.

Ask AVA"Help me map and automate an internal workflow for handling new customer orders."
🏢

5. Enterprise automations — finance, HR, onboarding

Live inside or around core systems — finance, HR, CRM, ERP, compliance. Often need stronger governance and IT involvement.

  • Finance workflows (purchase approvals, reconciliations)
  • HR onboarding/offboarding journeys
  • Compliance checks and audit-friendly trails

Typical tools: Microsoft Power Platform, Salesforce Flow, UiPath, Workato, Google Apps Script.

Ask AVA"We want to automate parts of onboarding and finance approvals. Where should we start?"
🤖

6. AI agents — multi-step reasoning & background work

More advanced — they can read information, reason across several steps, and act in the background continuously.

  • Monitoring inboxes and routing important items
  • Research or analysis workflows
  • Complex "if this, think about that, then decide" patterns

Typical platforms: Relevance AI, Zapier AI Agents, OpenAI Assistant workflows, custom builds.

Ask AVA"Help me understand if I actually need an AI agent, or if a simpler automation will do."
Ask AVA about your situation Seen enough to know what you need?

Frequently asked questions

These come up a lot. AVA can expand on any of them in conversation.

Does AVA actually build automations or agents for me?

No. AVA is a recommender, not a builder. It helps you understand what's possible, compare tools like Zapier, Make, Power Automate, n8n, and Relevance AI, and shape a sensible first version of a workflow or agent.

AVA does not deploy, host, or run automations — that's done in the tools it recommends, or by a person or partner you choose to work with.

Why does AVA keep mentioning Zapier, Make, or Power Automate?

Because those are the most practical starting points for most businesses — widely used, well supported, and compatible with hundreds of other tools without needing to write code.

If you're on a specific stack (Microsoft, Google Workspace, Salesforce), AVA will suggest native options like Power Platform, Google Apps Script, or Salesforce Flow instead.

What's the difference between a chatbot, a processor, an orchestrator, and an AI agent?
  • Chatbot: talks to people, answers questions
  • Processor: reads emails or documents and updates systems
  • Orchestrator: chains multiple steps into a journey
  • Agent: thinks across steps and can run in the background

AVA can help you work out which one fits your situation.

Do I need to understand LLMs or RAG to use this?

No. Most automations don't require any technical AI knowledge. Those details are handled behind the scenes by tools like Chatbase, Relevance AI, and OpenAI Assistants.

The glossary at the bottom of this page explains those terms if you're curious — but you don't need them to get started.

Do I need Microsoft, or does this work on Google / iPhone setups?

You don't need Microsoft. Many solutions work equally well with Google Workspace, Apple devices, and standalone tools like HubSpot, Mailchimp, and Shopify.

If you do use Microsoft (Dynamics, Business Central, Power Platform), AVA can help with that too.

Are there free or low-cost tools I can try myself?

Yes. Most platforms have free or trial tiers:

  • Automation tools (Zapier, Make, n8n, Pipedream) — light free plans for simple flows
  • No-code builders (Bubble, Bolt.new, Glide) — prototype apps without coding
  • AI platforms (OpenAI, Claude, Relevance AI) — experiment with agents and workflows

AVA can suggest where to start and what to try first.

What if I want someone to actually build it for me?

Once AVA has helped you understand what you need, you can take that to:

  • Your internal IT or automation team
  • A freelancer who knows the tools AVA recommends
  • A consultancy or partner for larger projects

AVA's output gives you a clear brief — so you're not starting from a blank page when you talk to someone.

Why doesn't AVA link directly to the tools it recommends?

AVA stays neutral and doesn't partner with or endorse any vendors. It gives you the plain domain name (e.g. "zapier.com") so you can find it yourself and make your own assessment. You stay in control of where you go next.

How your automations actually get built

Once AVA has helped you decide what you want, there are three sensible routes.

1. Self-build (no-code / low-code)

Use tools like Zapier, Make, Power Automate, n8n, Bubble, or Bolt.new to build simple flows yourself. Most use drag-and-drop builders.

  • Best for smaller teams and experiments
  • No deep coding required
  • Gives you hands-on learning

AVA can help you choose a tool and outline your first workflow.

2. With an AI/automation developer

Work with a freelancer or developer who already knows the platforms (Zapier, Make, n8n, Relevance AI, Power Platform, etc.).

  • Best when logic is more complex
  • You still own the process and tools
  • Use AVA's output as a mini-brief

AVA helps you describe what you need clearly, so you get better proposals and fewer surprises.

3. With a consultancy or partner

For bigger projects — design, build, and support from a specialist automation or AI consultancy.

  • Best for multi-team, multi-system projects
  • Helpful when change management is also needed
  • AVA helps you clarify scope before you talk to them

Use AVA to rough out the "what and why", then take that into conversations with potential partners.

Plain-language glossary — terms that come up around automation and AI

You don't need to memorise any of this. It's here if a term comes up that you don't recognise.

LLM (large language model)

The AI engine that reads and writes text. AVA runs on this kind of model.

RAG (retrieval-augmented generation)

A way for AI to look up your documents or data before answering, so responses are more specific to your content.

Vector store

A specialised way of storing information so AI can find related pieces quickly when answering questions.

Fine-tuning

Training an AI model on extra examples so it follows a particular style, tone, or set of rules more closely.

Agent

Software that can perform multiple steps, make decisions, and act in the background based on goals or rules.

API

A "plug socket" that lets one system talk to another automatically. Most automation tools use APIs behind the scenes.

Good starter questions for AVA

If you're not sure what to type, these work well. Copy and paste, or use them as inspiration.

Start using AVA Free to use — no account needed.