22 / CRM
RelateIQ
Now known as Agentforce
The problem
A CRM is only as useful as what people type into it, and salespeople do not like typing. Calls, emails and meetings happened all day, and most of them never reached the system. What did reach it was stale, and the tool said little about who to contact next. The people paying for that gap were the reps who lost track of a deal and the managers who could not trust the pipeline view.
RelateIQ set out to fix it by pulling in email, calendar and other interaction data automatically. That created a design problem of its own. Once the product knew a lot, it had to show it without burying people. We were a small team with no existing patterns to borrow, and every screen had to teach a new way of working while still looking like a tool a sales team could pick up in a day.
Success meant a person could sign up, see their own relationships already filled in, and know what to do next.
A product that guesses also has to earn trust. If the system suggested a follow-up and a person could not tell it apart from a task a colleague had assigned, they would stop believing both. Every place the product spoke for itself had to say so.
The approach
I worked on paper first and kept the team around the whiteboard, so most decisions were argued over a sketch before anyone built anything.
- 01
Onboarding on the whiteboard
I drew the first-run flow on a whiteboard with the team: sign in with email, create an organization, connect data sources, define a funnel, add a field. Red marker held the goals and the words we wanted people to read. The notes at the top, such as people do not read, shaped how little copy each step got.
Onboarding flow on the whiteboard. - 02
Lists as cards and a timeline
I sketched a contact list as cards beside a side panel with a timeline of activity. The margin notes record the decisions: rarely used list options hide in a dropdown, call to action buttons appear on hover, and the grid slides over the list sidebar when a card opens. I shared pencil pages like this with engineering before any pixels.
Hand-drawn wireframe. - 03
Pipeline filters
For the sales pipeline I annotated a sketch of the filter bar line by line. Status and owner always show because most processes depend on them. Other filters wait behind a More menu or a lightbox. Saved filters sit in a dropdown, and free text narrows results as you type instead of using checkboxes.
Annotated sketch of the pipeline view. - 04
Opportunities as a working table
The opportunities list became a working table: status, owner, inactive days, last communicator, lead source and next event date. Inactive days got its own column and its own color, since a deal going quiet was the thing a rep most needed to catch. The toolbar kept the saved view and filter controls from the sketch.
The opportunities list. - 05
The follow-ups list
Follow-ups combined tasks people assigned with ones the product suggested. I split the list into Today and Upcoming and labeled each row as assigned to you or suggested by RelateIQ, so a person could tell a machine suggestion from a colleague's request. A Send Email button sits on the suggested rows.
Follow-up suggestions. - 06
iPhone post-call flow
For the iPhone app I mapped what happens after a call: how a person starts one from a list, a follow-up, a notification or search, then what the app asks when the call or calendar event ends. The aim was a quick way to add a comment or a follow-up while the call was still fresh.
Early flow sketches for the iPhone app.
The solution
RelateIQ tracked relationships on its own. The web app showed a list of people or deals beside a conversation timeline built from email, calendar events and calls, with fields like status, owner and inactive days alongside. Pipelines and opportunities were views on the same data, and the follow-ups list suggested who to contact next and let people send the email from the list.
The idea that held the screens together was that the product should fill in the record and the person should only correct it. Each screen started from what the system already knew and asked for as little as possible. Typing was kept for the moments where only a person knew the answer, such as a comment after a call or a field added for their own process.
Decisions along the way
Paper before pixels
Flows started on the whiteboard and screens in annotated pencil, shared with engineering before anything was drawn on a computer. The reason for each choice was written in the margin, so it travelled with the sketch.
Status and owner always show
Most sales processes depend on status and owner, so those two filters never hide. Everything else waits behind a More menu or a lightbox, and free text narrows the list as you type.
A column for going quiet
Inactive days got its own column and its own color in the opportunities table, because a deal going quiet was the thing a rep most needed to catch.
Label the machine's suggestions
Each follow-up says whether it was assigned to you or suggested by RelateIQ, so a person can always tell the product's idea from a colleague's request.
What I delivered
- Whiteboard flows and onboarding sequence
- Annotated pencil wireframes for lists and contact views
- Sales pipeline and opportunities list designs
- Follow-ups list with suggested follow-ups
- iPhone app flow sketches and screens
See it working

Value added
For people using it
Reps saw their relationships and activity already in place, and the follow-ups list told them who to contact next. They typed less and could still correct anything the system had guessed.
For the business
The product stood apart from older enterprise CRM by cutting manual data entry and surfacing insight. A clear onboarding flow and focused pipeline views gave sales teams a quick way to see value at launch.
For the team
Sketches with written reasons in the margins gave engineering and product a shared record of why each choice was made. With a small team, that cut down on repeated debates.
By the numbers
What RelateIQ was valued at
Figures reported by TechCrunch at launch and in the year after. They describe the company, and none of them is a result claimed for the design.
View as table
| Period | Value | Note |
|---|---|---|
| Jun 2013 | $100M | At public launch, after $29 million raised |
| Mar 2014 | about $250M | A $40 million round |
| Jul 2014 | up to $390M | Salesforce agreed to buy RelateIQ |
Source: TechCrunch, June 2013 and July 2014 (opens in a new tab)
The result
RelateIQ launched publicly, after a long period of quiet testing, with native iPhone and Android apps. The company is now known as Agentforce.
What I took from it
Putting the reason for each decision in the margin of the sketch saved me from re-explaining it in every meeting. On the next product I would start the first-run flow even earlier, because everything else depends on what a new person sees first.
Background
RelateIQ opened to the public on June 12, 2013, after about two years of quiet testing. TechCrunch described it as a newer alternative to older enterprise CRM. It was meant to cut down manual data entry by pulling in email, calendar and other interaction data. It launched with native iPhone and Android apps. TechCrunch reported that it launched with about 100 clients and $29 million raised, priced at $49 or $99 per user per month, with free mobile apps. In March 2014 it raised $40 million at a valuation of about $250 million, and in July 2014 Salesforce agreed to buy it for up to $390 million.