Harry

[Harry @FounderFlows]

@yacineMTB The biggest question

What if?

[Harry @FounderFlows]

@DearS_o_n I disagree

[Harry @FounderFlows]

3/ You'll quickly spot patterns, priorities, and the language customers actually use.

Then talk to 2–3 customers about the biggest theme.

AI does the sorting. You do the understanding.

[Harry @FounderFlows]

1/ Your support inbox is a customer research database most teams never use.

Here's a simple AI workflow.🧵

[Harry @FounderFlows]

2/ Once a week (or month if volume is low):

Export your last 30–100 support tickets.

Ask ChatGPT or Claude to:

Group recurring problems
Pull exact customer language
Flag feature/process requests
Highlight severity and frequency

[Harry @FounderFlows]

3/ You're not copying competitors.

You're finding white space and sharpening your positioning.

Do it every month, not once a year. The compounding effect on messaging and product decisions adds up.

[Harry @FounderFlows]

1/ Competitor research is one of the highest-ROI uses of AI that many founders still treat as a quarterly task.

Do this instead.🧵

[Harry @FounderFlows]

Spend 15 min with any ai

• List your top 5–8 competitors (or provide your own)
• Extract for each:
Core positioning

Pricing (if public)
Target customer language
Content themes
Obvious gaps
Then ask:
"Where are they weak or silent in ways that matter to my ideal customer?"

[Harry @FounderFlows]

3/ Start with your own calls.

The feedback is often uncomfortable—and that's why it works.

You don't need an expensive sales-enablement platform to get better.

[Harry @FounderFlows]

1/ Most sales calls don't fail on the call. They fail after it.

The recording sits unopened.
The transcript goes unread.
The same discovery mistakes repeat.
🧵

[Harry @FounderFlows]

2/ After every call:

• Get the transcript.
• Paste it into Claude or ChatGPT.
• Ask for:
Pain points (using the customer's exact words)
Objections + how you handled them

Questions you missed
Competitor mentions
Next steps + owner
One coaching tip for your next call

[Harry @FounderFlows]

4/ Most founders save meaningful hours here before they ever need a sophisticated AI agent.
Start narrow. Measure time saved for two weeks. Then expand.

[Harry @FounderFlows]

1/ Two of the highest-ROI AI workflows for founders are still the unsexy ones:
• Inbox triage • Meeting intelligence
🧵

[Harry @FounderFlows]

3/ A simple setup:
→ Route and label emails → Surface only what needs a decision → Draft replies in your tone for approval
After meetings:
→ Extract decisions → Assign owners → Track deadlines in one place

[Harry @FounderFlows]

4/ If your AI agent feels fragile, the answer is often less autonomy—not a smarter model.

[Harry @FounderFlows]

1/ The largest look at AI agents running in production found something surprising:

The best agents are deliberately simple.
🧵

[Harry @FounderFlows]

3/ The pattern is clear:
Production teams are trading autonomy for reliability—on purpose.
The flashy demo is a digital employee.
The real production agent is a tightly bounded workflow with clear stop conditions and a human in the loop.

[Harry @FounderFlows]

@MartinGTobias https://t.co/cKGrIoc9dP