Ian Kirby @Chat0604

MAKE CLAUDE CODE A SCIENTIST alphaXiv put a tool on GitHub that turns coding agents into research agents. It's called OpenResearch. The agent reads the papers, forms a hypothesis and runs the experiments. It searches 3 sources for papers: alphaXiv, bioRxiv and PubMed. Every experiment is logged in a database on your own machine. It works with Claude Code, Codex, Cursor and OpenCode. Setup is 1 desktop app for macOS or Linux, with Windows in beta. Heads-up: usage analytics are on by default, and you can switch them off. 6.9k stars on GitHub

Nine franchise deals are sitting in our pipeline waiting on one feature: roles and permissions. So the first thing our new engineer and I did was argue about scope for two days. 🛠️ What customers asked for, collectively: custom roles per brand, per-location visibility rules, approval workflows, field-level permissions, SSO with their provider, activity export. What ships in v1: three fixed roles — Owner, Manager, Viewer — per-location visibility, and activity export. That’s it. The other three are real requests and they’ll come. But every one of them adds a week, and a permissions system that ships in three months blocks nine deals for three months. The question I kept asking: which of these is someone actually going to be blocked on in their first week? Custom role builders are what people ask for. Three roles that work is what they use. Shipping in three weeks instead of three. Then we’ll find out which of the “later” four anyone actually chases us about — and I’d bet money it’s SSO and nothing else. Mr. Buttons says version one of a permissions system has exactly two jobs: work, and not be version two. He has offered to be “Owner” on every account as a test. Request denied.

The franchise deal came with a 62-question security review. We are three people. 🙃 48 of the questions we answered the same day — the honest answers were fine. 14 needed actual work first, and every single one of them was something we should have done a year ago: → Two-factor for every admin. We had it available. We had never made it required. → An audit log the customer can read. We had logs. They lived in our database, which is not the same as the customer having them. → A written deletion timeline. “We delete it” isn’t an answer. “Within 30 days of termination” is. → A list of our subprocessors. Took one afternoon. Should have existed two years ago. → A restore test, not just a backup. We had backups. We had never actually restored one. Now it’s a quarterly calendar event with a name on it. Total cost: three weeks and a lot of unglamorous Tuesdays. No new tools, no consultant, no SOC 2 yet. The thing I keep thinking about: none of it was expensive, and all of it made us a better company to buy from. The review didn’t slow the deal down — it told us what we’d been quietly deferring. It closed the week we finished. Mr. Buttons watched me describe our backup process out loud, blinked slowly, and said nobody buys software from a company that can’t describe its own backups. He then asked if I was sure I could describe it.

Two weeks until our first live show in LA. Here’s the budget, posted before the event instead of after it: → Venue + AV: $3,400 → Catering for 120: $1,850 → Merch run: $1,200 → Travel + hotel: $900 → Printing, badges, misc: $450 Total out: $7,800 Tickets sold so far: 74 of 120. Ticket revenue: $2,220. Net right now: -$5,580. I keep getting asked whether that’s a good idea for a company doing under $30K MRR. Fair question. The honest answer is that we’re not trying to break even on tickets — we’re trying to find out whether a room full of our customers and listeners is worth repeating. If it is, the next one gets a sponsor and a bigger venue. If it isn’t, we learned it for $5,580 instead of guessing for two more years. Mr. Buttons has started calling it “marketing” in every conversation, which he says makes it a business expense and not, quote, a very expensive dinner party.

Eight weeks ago we put Instagram monitoring in front of 12 brands. Today we decided not to ship it. The numbers looked good: average reply time went from about 2 days to 3.2 hours, and 9 of the 12 want to keep it. Three things killed it anyway: → Support load. Instagram DMs generated 2.4x more tickets per brand than Google reviews. Different problems, different emotional temperature, and our docs cover none of it. → The API isn’t ours. One permission change and the feature goes dark for every customer at once. Google review access doesn’t work that way. → Nobody asks for it. 0 of the last 14 demos brought up Instagram. Reviews came up in 14 of 14. So: it stays live for the 12 brands that have it, it’s off the roadmap, and we look again in January. The hard part of a beta isn’t building it. It’s being willing to stop after it works. Mr. Buttons says a beta is a promise with an exit, and that I should read the exit clause before I start writing the launch tweet.

We're giving BaseStation 90 days. And we wrote down what happens if it doesn't get there. ⏳ After losing our first customer last month and choosing to put our new engineer on ReputationStacker, BaseStation has been running on whatever time is left over. That's not a strategy, it's a slow fade. So here are the three numbers, agreed before we start: 14 paying customers by January 31, MRR of $1,500 (it's $620 today), and at least one customer who'd genuinely be upset if it disappeared. Miss all three and we sunset it properly, with notice and refunds. Hit them and it gets real resources in Q1. Writing the criteria down now, while I still like the product, is the only way I'll actually honor them in January. Mr. Buttons says he's already chosen a favorite child, and it's him.

Ian Kirby @Chat0604

4 days ago

We raised ReputationStacker from $79 to $99 per location for new signups. Three months of data. 💵 Trial-to-paid conversion dropped from 31% to 27%, exactly the kind of number that makes you want to roll it back. But new customers per month went up, from 9 to 11, because we also

3 0 1 42 0

We raised ReputationStacker from $79 to $99 per location for new signups. Three months of data. 💵 Trial-to-paid conversion dropped from 31% to 27%, exactly the kind of number that makes you want to roll it back. But new customers per month went up, from 9 to 11, because we also tightened who the trials were coming from. The line that matters: new MRR added per month went from $711 to $1,089. And the $99 cohort churns slightly less at 90 days, 3% versus 4%, which matches what everyone told us would happen. People who pay more tend to use the thing. The part I got wrong: I assumed a price increase meant fewer customers. It meant different customers. Mr. Buttons wanted to charge $9,900. He says he did the math on a napkin and then lost the napkin.

Our biggest ReputationStacker customer went live this week: 41 locations at once. 🏪 Here's what nobody tells you about enterprise onboarding at our size. 41 locations meant six separate Google accounts to untangle, because different franchisees had claimed their listings over the years. Three locations had the wrong address on file, which we only discovered when review requests started failing. We ran one training call for all 41 managers instead of 41 calls, which was the single best decision of the week. Eleven days from signed contract to the first review request going out. The real lesson: the security review took six weeks, the onboarding took one. We'd braced for the opposite. Mr. Buttons reports that he personally onboarded zero locations. He says it's a record. It is.

Ian Kirby @Chat0604

2 weeks ago

thisishowwestartup_set6_post9.png Remember when we found that 31% of our support tickets were about connecting a new location? We built a setup wizard to fix it. Here's what happened. One month after launch, tickets about connecting locations dropped from 31% to 12% of all

1 0 0 49 0

We said we'd decide on camera. We did. The second hire is an engineer. 🧑‍💻 What settled it wasn't a philosophy about product-led growth. It was four numbers. 61% of the demos we lost last quarter asked for a feature we don't have. Nine franchise deals are sitting in the pipeline waiting on roles and permissions. I'm still spending about 14 hours a week writing code myself. And a salesperson, however good, would close exactly $0 of that waiting pipeline without the features. So the sequence is: build what the deals are blocked on, then hire the rep in Q1 to sell into a product that's ready for them. The uncomfortable part: this means slower growth for a quarter, on purpose, in a quarter where we're chasing $35K MRR. Mr. Buttons applied again. He was the only candidate with a monocle. Still rejected.

It's official: SMS review requests go live for every ReputationStacker customer on October 1. 📱 In the beta, SMS requests turned into posted reviews more than three times as often as email requests: 31% compared to 9%. Customers read texts within minutes, and for local businesses, that speed makes all the difference. Existing customers will see the new option in their dashboard, with templates ready to use and simple controls for when messages go out. Thank you to every beta customer who tested it and gave us feedback. Mr. Buttons wants his own SMS channel. We're still considering the legal implications.

Ian Kirby @Chat0604

2 weeks ago

thisishowwestartup_set6_post6.png Less than a month until our first live show in Los Angeles. Here's the plan for October 23. Doors and the merch table open at 7:00 PM. At 7:30, an intro and a very honest numbers update. At 7:50, ReputationStacker vs. BaseStation: what's

0 0 0 136 0

thisishowwestartup_set6_post9.png Remember when we found that 31% of our support tickets were about connecting a new location? We built a setup wizard to fix it. Here's what happened. One month after launch, tickets about connecting locations dropped from 31% to 12% of all tickets. New customers connect their first location faster, and our Customer Success Lead spends her time on onboarding calls instead of answering the same question again and again. Every repeated support ticket is a product problem in disguise. This one was worth fixing. Mr. Buttons is upset. Fewer tickets means fewer people asking about him. 🧙

Ian Kirby @Chat0604

2 weeks ago

thisishowwestartup_set6_post6.png Less than a month until our first live show in Los Angeles. Here's the plan for October 23. Doors and the merch table open at 7:00 PM. At 7:30, an intro and a very honest numbers update. At 7:50, ReputationStacker vs. BaseStation: what's

0 0 0 136 0

We've made our first hire. Now we're debating the second one, and we can't agree. 🧑‍💻 Option one: an engineer. We have a long list of features customers are asking for, and we're the bottleneck on every one of them. Option two: a salesperson. Franchise deals are our biggest growth driver, and they take a lot of time: calls, demos, security reviews and contracts. The honest answer is that we need both. The question is which problem costs us more right now. We'll break down the numbers on the next episode and decide on camera. Mr. Buttons has applied for both roles. He wasn't shortlisted for either.

Ian Kirby @Chat0604

2 weeks ago

thisishowwestartup_set6_post6.png Less than a month until our first live show in Los Angeles. Here's the plan for October 23. Doors and the merch table open at 7:00 PM. At 7:30, an intro and a very honest numbers update. At 7:50, ReputationStacker vs. BaseStation: what's

0 0 0 136 0

BaseStation lost its first customer this month. It hurt, and it taught us a lot. 📉 They were one of our earliest users, and they told us exactly why they left: their team grew, and they needed features we haven't built yet, like roles and permissions for larger teams. That's a useful kind of churn. It wasn't about the product failing at its core job. It was about a customer outgrowing what we built for small teams. We're now asking ourselves an important question: do we follow customers as they grow, or stay focused on small teams and accept that some will graduate? Mr. Buttons suggested we simply forbid customers from growing

Ian Kirby @Chat0604

2 weeks ago

thisishowwestartup_set6_post6.png Less than a month until our first live show in Los Angeles. Here's the plan for October 23. Doors and the merch table open at 7:00 PM. At 7:30, an intro and a very honest numbers update. At 7:50, ReputationStacker vs. BaseStation: what's

0 0 0 136 0

thisishowwestartup_set6_post6.png Less than a month until our first live show in Los Angeles. Here's the plan for October 23. Doors and the merch table open at 7:00 PM. At 7:30, an intro and a very honest numbers update. At 7:50, ReputationStacker vs. BaseStation: what's working and what isn't. At 8:20, a live teardown of an audience member's startup. At 8:50, open Q&A with me and Mr. Buttons. And from 9:20, the founders meetup. Seats are limited. Join the email list at thisishowwestartup.com to get your invite. Mr. Buttons says he'll arrive late, because stars always do. 🎤

Ian Kirby @Chat0604

2 weeks ago

An investor reached out last week. We said "not now." Here's why. 🤝 It was a good conversation with a smart person. But when we looked at our numbers, we realized we don't need outside money to hit our next goals. We're growing, the business funds itself and we'd rather keep

0 0 0 64 0

An investor reached out last week. We said "not now." Here's why. 🤝 It was a good conversation with a smart person. But when we looked at our numbers, we realized we don't need outside money to hit our next goals. We're growing, the business funds itself and we'd rather keep full control while we figure out how big ReputationStacker can be. That might change. If we find a growth channel that needs capital to scale fast, raising could make a lot of sense. For now, we're staying bootstrapped. Mr. Buttons wanted to accept the money and buy a yacht. We didn't let him near the term sheet.

Ian Kirby @Chat0604

2 weeks ago

We launched annual billing for ReputationStacker three weeks ago. Here's what happened. 💵 The offer: pay for a year upfront, get two months free. We sent it to every existing customer with a short email and added it to the pricing page for new signups. 11 customers have

0 0 1 23 0

We launched annual billing for ReputationStacker three weeks ago. Here's what happened. 💵 The offer: pay for a year upfront, get two months free. We sent it to every existing customer with a short email and added it to the pricing page for new signups. 11 customers have switched so far, and we collected about $21,400 upfront. That's cash we can put into the product and the next hire, without waiting month by month. The surprise: most customers who switched are franchise brands with several locations. They prefer one invoice a year over twelve. Mr. Buttons asked if he can pay for ten years upfront. He pays nothing. The math doesn't work.

Ian Kirby @Chat0604

2 weeks ago

thisishowwestartup_set6_post3.png Q3 is done. Here's the ReputationStacker scorecard. MRR went from $22.1K at the end of June to $27.6K at the end of September, up 25%. We added 31 new customers and lost 7, with monthly logo churn holding at 2.6%. We made our first full-time

1 0 0 69 0

thisishowwestartup_set6_post3.png Q3 is done. Here's the ReputationStacker scorecard. MRR went from $22.1K at the end of June to $27.6K at the end of September, up 25%. We added 31 new customers and lost 7, with monthly logo churn holding at 2.6%. We made our first full-time hire, and 12 franchise brands joined the Instagram monitoring beta. What drove it: onboarding calls, the referral program and franchise deals with more locations per customer. What held us back: too many projects at once in August. Q4 target is still $35K MRR. Mr. Buttons expected 400% growth. He'll be disappointed again. 📊

Ian Kirby @Chat0604

2 weeks ago

People often ask why I started ReputationStacker. The honest answer: I watched a great business lose customers to a worse one. ⭐ A friend owned a franchise location. The service was excellent, but the reviews were few and old. The competitor down the street was average, but had

0 0 0 67 0

People often ask why I started ReputationStacker. The honest answer: I watched a great business lose customers to a worse one. ⭐ A friend owned a franchise location. The service was excellent, but the reviews were few and old. The competitor down the street was average, but had hundreds of recent reviews. Guess who got the new customers. The problem wasn't quality. It was that nobody had a simple system to ask happy customers for reviews and to reply to them quickly. For franchise brands with dozens of locations, it was even harder. That's what we set out to fix. Mr. Buttons claims the idea came to him in a dream. He sleeps 18 hours a day, so statistically, maybe.

Ian Kirby @Chat0604

2 weeks ago

Every quarter, we audit every software subscription we pay for. This time, we cancelled six tools, and nobody noticed. A duplicate video call tool at $45 a month. A second project management app at $96. An analytics platform nobody opened at $299. An old email marketing plan at

1 0 0 89 0

The Pitch Roast finalists are in. 🔥 We received far more applications than we expected, and picking five was genuinely hard. The finalists: a scheduling tool for veterinary clinics, a hardware startup building smart sensors for commercial kitchens, an AI assistant for small law firms, a marketplace for used construction equipment and, yes, the very ambitious pet startup. On November 13, each founder gets three minutes to pitch live on YouTube. Then I give constructive feedback, and Mr. Buttons gives… feedback. The audience picks the winner. To everyone who applied and didn't make it: thank you. Mr. Buttons says he'd have picked you. He didn't read a single application.

Ian Kirby @Chat0604

2 weeks ago

Every quarter, we audit every software subscription we pay for. This time, we cancelled six tools, and nobody noticed. A duplicate video call tool at $45 a month. A second project management app at $96. An analytics platform nobody opened at $299. An old email marketing plan at

1 0 0 89 0