AI gives every founder in your niche the same answer.
What makes your business different is in your head. AI cannot use it until someone writes it down and gives it to AI as context. I do that writing.
I am Krishna Goutham. I head product and technology at a premium admissions consultancy, so I build for a founder-led firm every day. Before that I ran a team of about three hundred.
Tell me about itDoes any of this sound like your week?
- 01
Every decision still runs through you, so you cannot take a week off.
- 02
Work sits almost finished, waiting for your yes, because you still check everything your team sends out.
- 03
You know something is dropping somewhere, but you do not know where.
- 04
You still close most sales yourself. What you say on those calls lives in your head and your old chats, so the person you hired cannot say it the same way.
- 05
Every client call starts from zero, because nothing from the last one was written down.
- 06
You know your firm should use AI, but you do not know where to start, and you have no time to work it out.
Most service firms need a better brief or a small tool before they need an agent.
When you show me a piece of work that only you can do, it falls into one of three buckets.
- 01
A better brief.
The work is a judgment call, and a good brief lets someone else make it. What is missing is your standard, written down. A friend sent four thousand cold emails with no written standard behind them, and almost nobody replied. An hour of writing fixes this.
Example: a firm wrote its content and ads from guesses about what customers ask. The real questions were in its sales calls. One brief told AI to read the call transcripts and list the questions, grouped the way customers group them. It took an afternoon, and the guessing stopped.
- 02
A small tool.
The work repeats, the rule is clear, and someone still does it by hand. The fix is usually a form, a couple of fields or a daily list, and it takes a few days.
Example: a founder could not answer "who owes us this month?", though every date was already in the CRM. Two new fields in that same CRM now show the answer, and the team no longer asks for it in a chat.
- 03
A system.
The work passes through several people over weeks, so the system has to remember what happened, and a person has to approve each judgment call. Few firms need this, and it is where every vendor wants to start.
Example: a math-education company reviewed every lesson by hand against a long rubric, about an hour per lesson. Now eight AI reviewers check each lesson, and the editors approve or reject their notes on one board. By the company's own count, a review takes under ten minutes, and the editor still makes the call.
Most work in a firm fits one of these three, and I have built all three. Usually the answer is the first or the second. If I think you do not need me, I will tell you.
Everything we have built since late 2024.
Find work like yours. The ones we killed are in here too.
Who works on your business.
I ran teams in edtech for ten years. Since late 2024 I build software with AI.
- 2014 to 2024.
At a large edtech, I headed content, operations and curriculum for a team of about three hundred.
I ran teams in edtech for ten years. Since late 2024 I build software with AI.
- 2014 to 2024.
At a large edtech, I headed content, operations and curriculum for a team of about three hundred.
- 2024.
I left and started two ventures that year. Neither survived.
- Late 2024.
I started building software by directing AI. The first paid work came in 2025. By early 2026 I was fully booked and had hired two developers.
- December 2024.
Joined a premium admissions consultancy as a consultant.
- July 2026.
They made me Head of Technology and Product. I own their product lines and every system the firm runs on.
- 2024.
I left and started two ventures that year. Neither survived.
- Late 2024.
I started building software by directing AI. The first paid work came in 2025. By early 2026 I was fully booked and had hired two developers.
- December 2024.
Joined a premium admissions consultancy as a consultant.
- July 2026.
They made me Head of Technology and Product. I own their product lines and every system the firm runs on.
When your work needs a second builder, Venky joins. He ran a content studio for ten years and now builds agents for founders.
- 2013 to 2023.
Ran a content studio of about two hundred people at a large edtech.
When your work needs a second builder, Venky joins. He ran a content studio for ten years and now builds agents for founders.
- 2013 to 2023.
Ran a content studio of about two hundred people at a large edtech.
- 2023.
He left to build learning content with AI. A person still checks every lesson before students see it.
- Since then.
He built interactive textbooks and lessons for school maths, science and English, and a course that still sells.
- 2026.
He builds agents for founders. One puts a shortlist of tenders in a team's inbox every morning. Another drafts a reply for each new support ticket.
I also work with a few more consultants. Each has fifteen or more years of building products, running companies and using AI in their own business.
- 2023.
He left to build learning content with AI. A person still checks every lesson before students see it.
- Since then.
He built interactive textbooks and lessons for school maths, science and English, and a course that still sells.
- 2026.
He builds agents for founders. One puts a shortlist of tenders in a team's inbox every morning. Another drafts a reply for each new support ticket.
I also work with a few more consultants. Each has fifteen or more years of building products, running companies and using AI in their own business.
Two years of building taught me where success with AI comes from.
- Thinking. Knowing what you do and what you want.
- Feeding it in. The files, the prompts, the connectors, the setup.
- The model. Prompting it, knowing where it fails, catching it when it does.
So I do not plan or build until I understand your business almost as well as you do.
Where I can help you
Pick the one that sounds like you.
- 01
Build it.
You know what you want built. I ship it with you in weeks.
It can be a product you want to sell, a tool your team keeps asking for, or an app that is half built and stuck. It can also be the piece of work that still runs through you, like the proposal only you can write.
- I learn how your business works and what this build has to do for it.
- I write down what good looks like, so the build follows your standard.
- I build the smallest version that does the job, with a developer where needed.
- I hand it to your team and train them on it.
You get a working product, your standard in writing, and the next thing to build.
- 02
Grow it.
You know what you want to test. I get it live in weeks.
A new channel, a new offer or a launch you have planned for months. You know it is worth trying, but your week has no room to set it up.
- I learn who pays you today and why they chose you.
- I set up the test: the page, the messages, the tracking, and whatever else it needs.
- I run it with you and watch what comes back.
- I tell you what worked, what did not, and what I would try next.
You get a real answer from real buyers, in weeks.
Ask your own AI where AI fits in your firm.
I wrote down how I set up AI for a founder-led firm. Your ChatGPT or Claude reads it and applies it to yours. It is free and takes about five minutes.
Claude shows a warning for any chat opened from a link, like this one. Don't worry, it's safe. It only asks Claude to read what I wrote and apply it to your business.
- 1It asks you up to six questions about your firm, one at a time.
- 2It tells you how your firm uses AI today.
- 3It describes what would change in your week at each next step, from what you told it.
- 4It names the first thing to set up.