Under the hood · How Sidekick thinksNever wonder why the AI said that.

Every Sidekick conversation is run by eight AI specialists reasoning over one shared conversation: reading intent, weighing sources, planning hand-offs. No script, no decision tree: judgment at every turn. Here’s one conversation, thought through live.

8 specialized agents1 shared conversationGrounded, not guessed
rezolve.ai · Sidekick, reasoning traceThinking live
Employee ↔ Sidekick
Agent reasoning what the team is thinking, live

A recreated reasoning trace: the same thinking your admins can audit, decision by decision, in the Explainability view →

One assistant, a team of specialists.

Instead of one model guessing at everything, each decision in a conversation goes to a specialist built for that exact job, and all of them share the same conversation.

Specialized

Each agent is built and evaluated for one job: understanding intent, choosing a response, writing an answer, gathering ticket details, and so on.

Context-aware

One shared conversation flows through every agent, so decisions stay consistent from the first message to the last.

Matched, not one-size-fits-all

The system calls on different AI models for different jobs, choosing the one best suited to that specific decision.

The eight agents, at a glance

Each one owns a single decision in the life of a conversation.

1
Understand

Routing Agent

Understands & routes

Reads every new message: what's really being asked, and how they feel, then routes it to the right specialist.

asks itselfWhat is this person really asking, and how do they feel about it?
2
Route

Knowledge Agent

Selects the response

Searches your knowledge, then picks exactly one way to respond: not just an answer, the right kind of answer.

asks itselfWhat's the best way to answer here, not just an answer?
3
Respond

Synthesis Agent

Composes the answer

Writes a fresh answer grounded in your own knowledge base, and cites where it came from.

asks itselfWhat do our own docs actually say, and can I cite them?
4
Hand off

Follow Up Agent

Gathers what's missing

Requests the details a support ticket will need, and knows when it has enough to hand off.

asks itselfWhat's still unknown, and is this person losing patience?
5
Hand off

Ticket Agent

Ticket preparation

Turns the whole conversation into a clean, correctly-routed ticket the desk can act on.

asks itselfHow would the best agent on the desk write this ticket?
6
Respond

Catalog Agent

Service catalog

Guides customers through your organization's official service requests, field by field.

asks itselfWhich official request fits, and what can I already infer?
7
Hand off

Escalation Agent

Reads the reply

Reads how a customer responds to an offered next step: a yes, a no, or a hesitation.

asks itselfWhat did they actually mean by that reply?
8
Respond

Web Search

Live web fallback

Looks beyond your knowledge base for trusted sources, only when your organization allows it.

asks itselfOur docs don't cover it, am I allowed to look further?
1
Agent 1 · Understands & routes

Routing Agent

The entry point: reads every new message and works out what’s really being asked.

Reasoning“Worked yesterday, broken this morning: that’s an incident, not a how-to question. Nobody’s asked for a person yet, but the urgency is real: this is blocking their whole morning. Worth searching: certificate renewal, VPN client errors.”
  • The kind of request it is: a question, an action to take, small talk, or a clear ask to speak with a person.
  • Language, region & search terms: the language being used, the customer’s likely region, and the best terms to search for an answer.

Why it matters

Every other agent downstream builds on this first read, it’s the foundation the rest of the conversation is built on. It also watches for clear signals that a customer wants to be escalated, so that request can be honored immediately rather than getting lost in a search for an answer. This step runs first on every single message, so the assistant is always re-orienting itself to what the customer means right now.

2
Agent 2 · Selects the response · the judgment call

Knowledge Agent

Once your knowledge base has been searched, this agent picks exactly one way to respond.

Reasoning“Three sources match. The pre-approved reply is close, but it predates the new VPN client: a fresh answer written from the current guide, with a citation, serves this person better. The web isn’t needed: our own docs cover it.”

An already-approved answer

Uses a ready-made, pre-approved response word for word.

A freshly written answer

Hands off to the Synthesis Agent to compose one.

A guided walkthrough

Steps the customer through an automation or a service-catalog request.

A clarifying question

Asks the customer one focused follow-up before searching further.

A web search

Only if your organization has enabled it for that topic.

An honest “I don’t know yet”

When nothing available actually answers the question.

Why it matters

The outcome of this one decision determines which specialist handles the reply next, it’s the true fork in the road for every conversation.

Agents 3 & 8

Two ways to answer

Only one of these runs for a given reply, the Knowledge Agent decides which source fits.

3 Synthesis Agent

Composes the answer

Writes a fresh answer grounded in your organization’s own knowledge: help articles, policies, and approved content. Only speaks from what was actually found; used whenever a generated answer, rather than a pre-approved or scripted one, is the right fit.

8 Web Search

Live web fallback

Steps in when your own knowledge base doesn’t cover the question, and only for topics your organization has explicitly allowed it to search. Looks at trusted web sources, writes a grounded answer, and always includes the sources it drew from.

6
Agent 6 · Service catalog

Catalog Agent

Manages the full, guided conversation for your organization’s official service requests.

Reasoning“They said ‘same as my last one’ their request history fills in the monitor model, so I won’t ask again. One field left: which office. Then it’s ready to confirm.”

Offers, then asks

Offers the matching service, and asks for exactly the information it needs: nothing more, nothing invented.

Adapts as the conversation moves

Recognizes a change of mind, a topic switch, or a request for more detail, and adapts instead of getting stuck.

Confirms when it’s ready

Confirms once everything required has been collected, so the request is ready to submit.

Guardrail: stays inside your catalog

The services it can offer, and the exact fields it asks for, come directly from your organization’s own service catalog, never invented on the fly. Every request it produces is one your team actually defined and can fulfill.

Agents 4 & 5

From conversation to ticket

Once a support ticket is the right path, these two agents make sure it’s complete and well-formed.

4 Follow Up Agent

Gathers what’s missing

Once heading toward a support ticket, asks the follow-up questions needed to fill in whatever is still missing, and recognizes when enough has been gathered. Also watches for frustration, so a customer who’s had enough can be offered a person instead of more questions.

5 Ticket Agent

Ticket preparation

Takes everything gathered and writes a clear subject and description, with the correct category and routing from your own ticket taxonomy. A nice touch: before opening a new ticket, it checks whether a service-catalog request would serve the customer better.

7
Agent 7 · Escalation interpretation

Escalation Agent

Three matched specialists, one shared job: understanding how a customer responds to an offer.

Reasoning“‘Sure, if it’s quick’ isn’t a plain yes, it’s a yes with a time worry. Connect them now, and pass the note along: keep it brief.”

Offered a support ticket

Reads whether the customer wants one raised: including hesitation or a change of mind, not just a plain yes or no.

Offered a live chat

Reads whether the customer wants to be connected to a person right now.

Offered a choice of the two

Reads which path the customer picked when both a ticket and a live chat were on the table.

Why three specialists instead of one

Each offer calls for reading the reply in a different context, so rather than one generic yes/no check, each situation gets a reply-interpreter built specifically for it, and a response written to match.

How they work together

A roundtable, not an assembly line.

All eight agents sit around the same live conversation: every message, every detail gathered, every decision made so far. There’s no fixed pipeline: the specialist the moment calls for takes the floor.

One shared conversation

Every message, every detail gathered, every decision so far: read by all eight, all the time.

1Routing Agentunderstands & routes
2Knowledge Agentselects the response
3Synthesis Agentwrites & cites
4Follow Up Agentspots what's missing
5Ticket Agentcrafts the hand-off
6Catalog Agentguides requests
7Escalation Agentreads replies to offers
8Web Searchlooks further, if allowed

Nothing is repeated

What one agent learns (a detail, a failed fix, a mood), every agent knows. The employee never says anything twice.

The floor changes hands

A question can become a request, then a ticket, mid-conversation. Whoever’s specialty fits the moment takes over, with the whole story.

Judgment, not scripts

No decision tree. Each turn, the team re-reads the conversation and decides what this moment actually calls for.

Under the hood

Matched to the job

Different decisions call for different strengths, so the system doesn’t force one model to do every job.

Fast, precise classification

Understanding intent and choosing a response path are quick, structured decisions, handled by models built for speed and consistency.

Nuanced writing

Composing a clear, well-grounded answer calls for a model strong at natural, careful writing.

Careful reasoning

Reading a customer’s reply, gathering ticket details, and structured hand-offs draw on models suited to step-by-step reasoning.

The system genuinely draws on more than one leading AI provider, choosing the model best suited to each individual decision rather than routing everything through a single one-size-fits-all model. That choice is an implementation detail your team can tune per agent as models improve, the eight decision points above stay the same.

Built in, not bolted on

The same principles, every agent, every time

Being built from specialists doesn’t mean being unpredictable, every agent follows the same ground rules.

Answer from what’s actually there

The Synthesis Agent and Web Search only speak from documents or sources that were actually retrieved, never open-ended invention.

Stay inside defined boundaries

The Catalog Agent and Ticket Agent only offer services and categories your organization has actually defined.

Ask rather than assume

When information is missing, the assistant asks a focused follow-up instead of guessing at an answer or a ticket’s details.

Know when to hand off

Clear signals (an explicit request, frustration, or a stalled conversation), route a customer to a person rather than looping.

One shared conversation

Every agent works from the same context, so a customer is never asked to repeat something they already said.

Honest about not knowing

When nothing available answers the question, the assistant says so, rather than presenting a guess as a confident answer.

The takeaways

Your assistant isn’t one model guessing.

Eight specialists, one conversation

Each agent owns a single decision, and all eight share the same context, all the time.

Every reply is a judgment call

Every possible way to respond gets weighed before a single word is written.

Grounded by design

Answers come from what was actually found: cited, never invented.

Built to hand off

Tickets, service requests, and live agents are first-class outcomes, not fallbacks.

Matched to the job

Different decisions draw on different AI models, chosen for what each does best.

“Not one model guessing: a team of specialists, each doing one job well, sharing one conversation.”

See it run on your own tickets →

Same eight agents, your knowledge base and service catalog, book a live demo.

Book a demo