Fixholics Talk to us
Fixholics Assistant · Agentforce agent

An agent thatonly says whatit knows.

A customer-facing Agentforce agent for fixholics.com. It answers questions about the company, the services, how engagements work and the case studies, using a fixed set of facts and nothing else. When a visitor wants to talk, it collects their details and creates a Lead through an invocable Apex action.

  • Built with Agent Script, Nov 2026
  • Lead capture: invocable Apex with tests
Our Agentforce practice Build one for us
A conversation

Helpful, grounded,and honest about limits.

The exchange on the right is what the agent does with three ordinary questions: answer from the facts, refuse to invent a price, and hand the visitor to a human with their details saved.

  • Welcome message and a router that always picks exactly one subagent
  • Subagents for about-Fixholics, ambiguous questions, off-topic requests and escalation
  • Casual tone, plain language, English by default
Fixholics Assistant
Do you work with companies outside India?
Yes. Fixholics is India-based and works with clients worldwide, across time zones. We agree overlap hours at kickoff and keep every decision in writing.
Roughly what would a CPQ to Revenue Cloud move cost us?
Fixholics doesn't publish prices, and I won't guess a number. The team gives an estimate after a short conversation about your org. Would you like a senior consultant to get in touch?
Yes. I'm Priya Nair, priya@example.com.
Thanks! Your details are saved and a senior Fixholics consultant will reply within one business day.
What it answers

Four topics.One source of truth.

Each topic is scoped in the agent spec, with the facts it may use written into the system instructions.

01

About Fixholics

Who we are, founded 2017, how engagements run, the four stages and the principles.

02

Salesforce solutions

Every service we offer and what each covers, from Agentforce to managed services.

03

Industries served

Telecom and media, SaaS and technology, professional services, manufacturing.

04

Contact

Collects name and email when a visitor wants to talk and saves them as a Lead.

How it is built

Agent Script,one Apex action.

The whole agent is source in a Salesforce DX project, published with the Agentforce CLI and snapshotted to GitHub so it can be rebuilt in a fresh org from a seven-step README.

@InvocableMethod(
    label='Create Fixholics Lead'
    description='Saves a website visitor who wants to talk to the Fixholics team as a new Lead.'
    category='Fixholics'
)
public static List<Response> createLeads(List<Request> requests)
  • Bulkified invocable Apex, with sharing, CRUD and FLS enforced, duplicate rules respected, errors returned never thrown, with its own test class
  • A permission set (Fixholics_Agent_Lead_Capture) assigned to the Einstein Agent user, nothing broader
  • Agent Script bundle: system instructions, router, subagents and linked variables for the messaging session
  • Settings deployed as metadata (Bot, Agent Platform, Einstein Copilot), so the org is reproducible
Questions

About theassistant.

Where does the assistant run?

In the Fixholics Agentforce org, published and activated through the Agentforce CLI, where it answers with live actions. The same bundle deploys to any org with Agentforce through the rebuild guide in its repository.

How do you stop it making things up?

The system instructions carry the only facts it may use, and tell it never to invent prices, budgets or timelines and never to confirm which countries clients are in. Questions outside those facts are routed to an off-topic or ambiguous-question subagent that asks for clarification or hands over to a human.

What happens when someone wants to talk?

The router hands the conversation to a lead-capture flow that collects the visitor's name, email and question, then calls an invocable Apex action that creates a Lead with source Web. It runs with sharing, respects CRUD, FLS and the org's duplicate rules, and returns an error message instead of throwing.

Can you build one for our customers?

Yes. The same approach applies to a customer-facing agent on your own data: ready the data and permissions, write the facts and routing, build the actions on Flow or Apex, then test in the Testing Center before it goes live.

Also by FixholicsAll products
FixShiftSalesforce app · 3 packages + CLI FixLensChrome extension · v2.0.0 fixyInspectorChrome extension · v1.0.0 FixTraceChrome extension · v1.0.0 FixyOfficeOffice suite · Web, Windows, Android, iPhone, iPad, Mac FixyBillBookBilling app · Web, Windows, Mac, iPhone, iPad, Android fixyEditPDF toolkit · Web, Windows, Mac, iPhone, Android FixholicsApp · iPhone, iPad and Android FixyTaskApp · iPhone, iPad, Mac and Android ShubhVelaApp · iPhone and Android · 12 languages FixyMindApp · iPhone and Android States Civic DeskApp · iPhone, iPad, Android and web · English and Hindi Fixholics Labs4 reference orgs on GitHub

An agent for your customers?

We start with a readiness review of your data, permissions and processes, pick one use case with clear value, then build, test and govern that agent before scaling.

Start with a readiness review Email us