As a SaaS founder you are the copywriter, the support team, the analyst, and the person who is supposed to be building. Every release needs a changelog and a few lifecycle emails. The landing page needs a rewrite every time positioning shifts. A competitor ships something and you want a clear-eyed teardown before lunch. The usage export sits in a CSV that you keep meaning to actually read. It is all real work and it all competes with shipping.
An AI agent is a good fit because so much of this is text and tabular data built on things you already understand. The agent runs real Python on an uploaded CSV to tell you who is about to churn instead of you eyeballing a spreadsheet, searches the live web to read a competitor's pricing and docs, and remembers your product, your ICP, and your voice so every draft starts from your context instead of a blank one. Because the agent runs on whichever model you pick, you can also ask Claude and GPT to phrase the same positioning line and see which one lands.
Put your product facts and voice in Memory once, then the prompts below cover the writing, the support scaffolding, and the analysis you keep postponing. You stay close to the product. The agent handles the copy and the spreadsheet.
Capabilities this leans on
Do this first. Copy, emails, and teardowns all improve once the agent knows your product, your ICP, and how you talk.
Remember these facts about my company: we make Tideline, a lightweight analytics tool for indie mobile app developers, at tideline.app. Pricing is Free, Pro at $29/mo, and Team at $99/mo. Our ICP is solo and small-team app developers doing $1k to $50k MRR who find Mixpanel and Amplitude too heavy and expensive. Our voice is direct, technical, and friendly, we talk to developers like peers, no buzzwords, no 'empower'. Main competitors are Mixpanel, Amplitude, and PostHog. Our edge is setup in under 10 minutes and flat, predictable pricing.
Get hero, subhead, and feature copy, then see two models take the same swing.
Write the hero section for tideline.app: an H1 under 8 words, a one-sentence subhead, and three feature blocks each with a 4-word heading and a 20-word benefit line. Speak to indie app devs who think Mixpanel is overkill. Direct and technical, no buzzwords.
Give me three alternate H1s with different angles: speed of setup, predictable pricing, and 'analytics that fit your size'.
Take this positioning line, 'Tideline is analytics built for the size you actually are,' and show me how you'd phrase it. Then switch the underlying model to GPT and ask it to phrase the same line, so I can compare the two side by side.
What you get: A full hero you can ship plus a real read on which model's phrasing fits your voice.
Paste the shipped changes; get user-facing notes in your voice.
We shipped these this sprint: custom funnel steps, CSV export on the Pro plan, faster cold-start on the SDK, and a fix for the iOS 17 session bug. Write a changelog entry in our voice: a one-line summary, then a short bullet per item written for the user, not the engineer.
Now write a 90-word in-app announcement highlighting custom funnels, and a 40-word version for the @tidelineapp social post.
What you get: Release notes, an in-app banner, and a social blurb from one paste, all sounding like you.
Draft the sequences that turn signups into Pro upgrades.
Write a 4-email onboarding sequence for a new free signup: day 0 welcome with the one thing to do first, day 2 'install the SDK' nudge with a link, day 5 first-funnel walkthrough, day 8 a soft nudge toward Pro. Each under 120 words, direct and developer-friendly, one CTA each.
Write a win-back email for a free user who hasn't logged in for 21 days, under 100 words, that asks what got in the way and offers a 15-minute setup call.
Save the onboarding sequence as a Skill called 'Lifecycle emails' so I can adapt it for the Team plan later.
What you get: A full lifecycle sequence ready for your ESP and a Skill you can re-run for other segments.
Cover the questions that flood the inbox after a launch.
Write eight support macros in our voice for the most common Tideline questions: SDK not sending events, how to set up a funnel, difference between Free and Pro, how billing proration works, GDPR and data residency, how to export to CSV, how to invite a teammate, and how to cancel. Keep each under 90 words and link where a doc would go.
A user wrote: 'events show up in debug mode but not in production.' Draft a calm troubleshooting reply that lists the three most likely causes in order and asks for the one detail you'd need next.
What you get: A starter macro set and a sharp troubleshooting reply you can paste into the helpdesk.
Live web search reads their pricing, docs, and reviews so you get a real picture.
Search PostHog's current pricing, free-tier limits, and what their docs say setup takes. Compare it to Tideline in a table on price, setup time, and target user, and tell me where we win and where we honestly lose.
Read recent PostHog and Amplitude reviews on G2 and Reddit and pull the five complaints that come up most. Which of those are things Tideline genuinely does better, and draft one comparison-page paragraph for each that's true and not slimy.
What you get: A sourced teardown and honest comparison copy you can put on a competitor page.
Upload the CSV; Python does the analysis you keep postponing.
I'm uploading a CSV with one row per account: signup_date, plan, mrr, last_active_date, events_last_30d, seats. Tell me how many Pro accounts are at risk, defining at-risk as no activity in 14+ days or events_last_30d down more than 50% versus the prior 30, and list them with the signal that flagged them.
From the same file, what's the difference in 30-day activity between accounts that upgraded to Pro and those still on Free? Chart it, and tell me the one behavior that best separates the two.
Every Monday at 8am, remind me to upload the latest usage export and re-run the at-risk analysis, and summarize what changed since last week.
What you get: A ranked at-risk list, the behavior that predicts upgrades, and a weekly habit to keep watching it.
Content marketing
Draft long-form, atomize it everywhere, and compare models to pick the best line.
Customer support
On-brand replies, reusable macros, thread summaries, and a daily triage digest.
Data & analytics
Upload a CSV and get real computed numbers, charts, and a weekly report, not guessed math.
Open the Agent, paste any prompt above, and change the details to fit your business.