Unlocking a Partner's First Agent

Partners are often excited about AI agents in the abstract but freeze when asked "okay, but what would you build?" Here's a practical framework for helping them find their first use case:

Once you've surfaced a few agent candidates, help them prioritize using two simple axes: how much time/pain does this cause? and how straightforward is it to automate? High pain + high straightforwardness = start here.

Set the right expectations for "first"

The biggest mistake is pushing partners toward something ambitious. The goal of the first agent isn't to transform the business — it's to build confidence and intuition. A small win that works reliably is worth infinitely more than a grand vision that stalls in scoping. Help them aim for something they can ship in weeks, not months.

⁠

Steps

Go through a few of the discovery questions. Always start with the Simple Workflows first

Create a shortlist together

Use Hector’s agent “What agent to build?”

1st prompt is the partner’s URL, + I’m meeting with this partner, help me brainstorm....

two agent ideas

an agent that does x, y and z

you will be asked about the primary end-user persona for these agent(s)

you will be asked if your partner has a public API or MCP server

2nd prompt - Does [partner’s main product] have the capability to do the [agent name suggested]

3rd prompt - Does [partner’s main product] have a feature that is similar to the [agent name suggested]

Discovery Questions

Simple Workflows

What are the top 1-2 things that a customer has to leave [key product] to go do?

What information do they copy-paste into or from [key product] on a regular basis?

Start with pain, not technology

The worst way to find a use case is to start with "what could an AI agent do?" The best way is to ask partners to describe their most tedious, repetitive, or error-prone workflows. A good first agent almost always lives inside a process someone already hates doing manually.

Prompts that unlock this:

"What does your team do every day that feels like it shouldn't require a human?"

"Where do things fall through the cracks between systems or people?"

"What would you do more of if it weren't so time-consuming?"

⁠

Use the 3 R's filter

A great first agent use case is typically:

Repetitive — it happens frequently enough to justify building it

Rule-based enough — there's a clear enough pattern that an agent can handle most cases without constant human judgment

Relatively low-risk — mistakes are recoverable, so the partner can experiment without fear

Things like summarizing meeting notes, qualifying inbound leads, drafting follow-up emails, pulling together weekly reports, or answering common user/customer questions tend to score well on all three.

⁠

Map it to a role

Different partner types have different natural entry points:

Sales → lead qualification, proposal drafting, CRM data hygiene

Managed service providers → ticket triage, customer onboarding, monitoring alerts

System integrators/consultants → research and discovery summarization, status reporting, RFP response drafting

ISVs → customer support deflection, onboarding flows, release note generation

EDU - Role

⁠

A "day in the life" exercise

Have customers walk through a single workday hour by hour and flag every moment they switch between tools, copy-paste information, wait for someone else, or do something they've done the exact same way 50 times before. Those friction points are almost always agent opportunities.

User friction points where Go Writing Agent integration adds value

Finding the context-switching pain

What does a user have to open a second tab to go do while they're writing?

What information do they copy-paste into emails or docs on a regular basis?

What do they have to look up before they can finish a sentence?

Finding the writing/rewrite angle

Is there a 'right way' to write something in your product's domain — a tone, a format, a structure — that most users get wrong?

Do your power users write differently than new users? What's the gap?

What do your best users do before they hit send that others skip?

Finding the data angle (knowledge layer)

What data lives in your product that a user wishes they had in front of them while writing?

What do users search for in your app right before switching back to their email?

Finding the action angle (MCP/formulas)

What's a one-click action in your product that users currently have to navigate 3 screens to do?

What do users ask your support team how to do that's actually just a single API call?

Validating the persona fit

"Who in your customer base complains most about repetitive work?"

"What does your most productive user do in the first 5 minutes of their day?"

The best questions are the ones about copy-paste habits and tab-switching — those are almost always a direct map to a Go Agent use case.

Technology Stack Mapping Exercise

Map out the top 5 apps/tools that a persona uses and the gaps, task switching, and copying and pasting that happen on a frequent basis