←
Avinash Saurabh106 posts
Avinash Saurabh avatar

Avinash Saurabh

@xavinash

Built and sold 2 VC backed ventures. Now bootstrapping and building in public. Sharing on Starting up, GTM, AI. deepsmith.ai aeonut.com meetspindle.com

◷ Joined November 2025▧ 20 Photos and videos
Avinash Saurabh avatar
Avinash Saurabh@xavinash·1h

James Currier explained how a startup beats better-funded competitors. The field manual: He was CEO of four venture-backed companies, then co-founded NFX, which has raised $1.5 billion. NFX had a list of 48 things it looked for in founders. All of them rolled up into one: how fast things happen around them. How he says to get faster: 1. Set a time, not an output. Ask "what can I do in 30 days?", never "how long will my dream product take?" 2. Plan several launches at once. He sat a team he thought was going way too slow at a whiteboard for 90 minutes: five products, five names, one launch every 30 days. The second one, House Party, was the one they sold. 3. Ship version one ugly. "Speed is quality." Version six will beat the perfect version one you would have built in the same time. 4. Launch away from your ego. If the company is called Color, don't launch on color.com. 5. Borrow, then scale. Google copied search and Airbnb copied VRBO, then charged 10% instead of $300 a month. The full manual is in the images.

Post attachmentPost attachmentPost attachment
Avinash Saurabh avatar
Avinash Saurabh@xavinash·2h

Some places exist only for the very rich, and most people have no idea they're there. @grok which 5 are they, and what does it really cost to get in? One number each.

Avinash Saurabh avatar
Avinash Saurabh@xavinash·3h

A rule worth keeping when you put AI into a workflow: 70% accurate is not 70% automated. The CEO of Leo, which builds procurement agents, says his company's first product is now an 8-hour exercise for engineering candidates. A customer could build it too and reach maybe 70% of the performance. But 70% performance doesn't mean 70% automation. People still have to check all of the data, so you may have made more work than before. I think this is why a good demo tells you so little. The demo shows the 70%. The cost is in checking every output. The rule stops where nobody did the work before. Big companies, he says, mostly pay anything under $50k without negotiating, because nobody has the time. There, a weaker agent still beats nothing.

Avinash Saurabh avatar
Avinash Saurabh@xavinash·4h

How to take a new sales rep from day one to live calls in 14 days. This is the bootcamp Alex Hormozi's team runs for every new hire. They train apart from the rest of the team, so nobody buries them in tips. 1. Day 0, the weekend before: a one-on-one, and homework due Monday. 2. Day 1: sell them the process before you teach it. The values, the path, the ramp and the gates they will have to clear. 3. Days 2 to 4: one part of the script a day. Day 2 is the intro and the early brush-offs ("I don't have time"). Day 3 is discovery, the heaviest day. Day 4 is the move into the pitch, the pitch itself, and booking the next call. 4. Every day: they read the script out loud 25 times (about 2 hours) and get three one-on-ones with the manager, at the start, middle and end of the day. Role-play against easy prospects with scripted answers and no curveballs until the whole script is automatic. 5. Day 5: the full script under pressure. Day 6: their first real dials. 6. Day 7: script graduation. The manager passes or fails each rep on script adherence, tonality, recaps and handling the top five objections. 7. Day 8: ask each rep to DM you what to start, stop and continue. Privately, because the first answer given in a group anchors everyone else. If more than three reps say the same thing, change the playbook for the next group. 8. Days 8 to 14: more time on the phones, shadowing live calls, and reviewing recordings of their calls. Then two gates before they get a full calendar. Gate 1: a cap of 30 live connects a day. That should give 10 conversations of 3 minutes or more and 2 to 3 booked meetings. They hold that for a week. Gate 2: a cap of 3 booked meetings a day until the show rate clears 60%. Tell them on day one that it is a performance seat with 60 days to prove it, so nobody is surprised on day 45.

Post attachment
Avinash Saurabh avatar
Avinash Saurabh@xavinash·5h

Some ad slogans stay in people's heads word for word for decades. @grok rank the 5 stickiest slogans of all time, and the one thing each did that most taglines don't. Best first.

Avinash Saurabh avatar
Avinash Saurabh@xavinash·5h

Likes don't pay rent. Drop your URL and let real people try what you made.

Avinash Saurabh avatar
Avinash Saurabh@xavinash·6h

How to register your business on ChatGPT and see how often it recommends you. The feature is called Businesses in ChatGPT, and it is in beta. Setting it up takes a few minutes. Open ChatGPT and go to Business Hub. Search for your business. Verify it with a code sent to your business phone or email, a DNS TXT record, or a Shopify login. Fill in your description, hours and contact info. Add your products if you sell any. You now have a verified profile. When ChatGPT mentions you, people can tap through to it. Open the analytics dashboard. It shows four things: How many times ChatGPT mentioned your business. Website clicks from ChatGPT. Profile views. Your top 5 discovery topics, which are the questions people asked when you came up. Read the discovery topics first. They are a free list of the questions you're already winning. Then ask ChatGPT each of those five questions yourself and note who else it names. Those are the businesses you are up against for that answer. Online and software businesses can claim a profile from inside or outside the US. Local shops and restaurants can only claim existing US listings for now.

Post attachmentPost attachment
Avinash Saurabh avatar
Avinash Saurabh@xavinash·7h

Fastlane put one button in its app: "Book a call with the team if you want 7 days of extra access." Just over 2 months after launch, it makes $69K a month in recurring revenue, its co-founder says. That button got them to 2,000 customer calls. What they did with the calls: 1. They kept a weekly quota. He says he and his co-founder have done around 20 calls a week, every week, since customer discovery began. 2. They traded access for time. The button had a rainbow border and opened a Calendly page. Users booked because they wanted the extra week. 3. Discovery calls followed The Mom Test. They never asked "would you use this?" They asked: walk me through the last time you solved this problem. How much time or money did it cost you? What would happen if you did nothing about it? 4. Usability calls were silent. Every waitlist sign-up got an email inviting them to a call. On the call, they shared their screen and used the product while the founders watched and said as little as possible. 5. Success calls went to power users. They tracked usage, called the heaviest users and asked how the product was helping them. Views, conversions and app installs from short-form video came up, so they doubled down on that. 6. Every call went into one database. An AI note taker on each call, the notes in Notion, all of it plugged into Claude as a customer dashboard with a "customer love score". Their non-technical co-founder built it with Claude Code. It showed that people who signed up out of curiosity retained worse than those who picked "I need marketing now". 7. The calls narrowed the product. The first version tried to do SEO, Reddit and short-form content in one tool. They went all in on short-form, shipped a beta to a 2,000-person waitlist, and ended up with a swipe-right screen that works like Tinder. He says they are past 1,000 paid users now. The reason it worked, I think: the button made every call cheap for both sides, so the calls never stopped.

Post attachmentPost attachment
Avinash Saurabh avatar
Avinash Saurabh@xavinash·10h

The last Baby Boomers are leaving the workforce, and some everyday things are leaving with them. @grok which 5 disappear first (habits, businesses, ways of doing things), and roughly when? Be specific.

Avinash Saurabh avatar
Avinash Saurabh@xavinash·11h

TitanX finished 2024 at $1.4M and 2025 at about $8.79M, its founder says. It began as a tool his sales agency bought so competitors couldn't use it. How an agency turned into a software company: 1. He bought the tool he was reselling. His agency used and resold a slow, expensive service that knew who was likely to answer the phone. When competitors started using it, he bought the company in 2023 and kept it for his own clients. 2. He listened when clients asked for the tool. About six months in, clients (Sony among them) wanted it for their own sales teams, and the agency had to say no. In his words, "the moat that's protecting the castle is more valuable than the castle it's protecting." 3. He shut the services business down. A seven-figure agency, plus another company spinning off about $500,000 a year in cash flow. He finished every client contract and put about $900,000 of his own money in during 2024. 4. He got the first version built for equity. A client who ran an AI development shop came in as fractional CTO on a vesting schedule. The build started in March 2024 and launched on June 1. 5. He asked existing clients what they would pay. One was asked about $50,000, the next about $30,000. A couple of them became the first customers. 6. He presold 100 beta spots. $199 for life: 1,000 credits and a credit price locked in forever. The first 100 sold in one day. The next 100, at $297, filled the following month. 7. He dropped self-serve after about a month and a half. TitanX sells to mid-market and enterprise sales teams that believe in the phone. 8. He sold one number. An average connect rate of 25%, against 3 to 6% in the US, he says. He named the category phone intent and has campaigned against high-volume dialing for years. The reason it worked, I think: he had been the customer first. He used the tool, resold it and ran it for clients before it was software, so on launch day he already knew what it delivered.

Post attachmentPost attachment
Avinash Saurabh avatar
Avinash Saurabh@xavinash·12h

You run a 3-person business and pay for a dozen software subscriptions. @grok name the 5 free tools that could replace the most of them, and what each one replaces. Be specific.

Avinash Saurabh avatar
Avinash Saurabh@xavinash·12h

Which AI product are you most proud of this month? Drop the link and let's see what's next.

Avinash Saurabh avatar
Avinash Saurabh@xavinash·13h

A lot of sales teams are using AI to send more written outreach. I think the more useful question is what all that writing does to the inbox it lands in. Vanta's CRO says their main way of generating pipeline is now cold calling, with SDRs paired one-to-one with AEs. Her reason: "there's so much written slop." She calls the return on those calls unbelievable. When a written message costs nothing to send, it stops telling the reader that anyone made an effort. A live call still costs a person's time, and the buyer can hear that. If your outbound is all email and the replies have gone quiet, I'd test the phone before writing a better sequence.

Avinash Saurabh avatar
Avinash Saurabh@xavinash·14h

Selling to founders? I read every link below. If it fits what I need, I'll buy it.

Avinash Saurabh avatar
Avinash Saurabh@xavinash·15h

$100k a year without burning out. @grok which job gets you there with the least stress, and what does it really pay 2 years in? One job, one number.

Avinash Saurabh avatar
Avinash Saurabh@xavinash·15h

Ulrik Lehrskov-Schmidt's firm, Willingness to Pay, has run more than 200 B2B pricing projects. He explained how to start charging for AI without wrecking your base. The field manual: 1. Don't raise everyone 10 to 30%. Find the accounts where you already deliver far more value than you charge for. He says almost every company has a 50K account that would pay a million. 2. Launch the AI feature free. If you charge on day one and nobody adopts it, you can't tell whether the product failed or the price did. 3. Then follow his sequence: guardrails, free launch, adoption, learn how customers use it, meter it, charge. Figma did this and found what he calls a "double unlock": credits captured the value of the AI, and the value the seat model had been hiding. 4. Put one person in charge. "You need to have a pricing dictator." Two similar clients rolled out the same kind of credit system. One took nine months, the other four weeks, and the difference was org layers. 5. Treat price as the budget for everything else. Charge 200 instead of 100 and you can spend twice as much on acquisition and product. Jason Lemkin argued the other side: charge from launch, and if it doesn't sell, take that as the market's answer.

Post attachmentPost attachmentPost attachment