WITH SANTOSH GHOSAL
Before the First AI Client Says Yes
Santosh Ghosal wanted to start an AI consultancy in four weeks. He came looking for the right team, pricing and technology. The hour brought him back to the person who would have to trust him enough to buy.
An enterprise project manager considering life as a co-founder, Santosh asked how to start, deliver and sustain a consultancy that works inside a customer's business. As the room worked through those questions, customer value and the human side of change became the starting point.
Listen onApple Podcasts Spotify
OCT03Santosh
Ghosal
EXPLORING AN AI CONSULTANCY
A four week launch × A first customer
WHAT EARNS THE FIRST YES?A company on a four week clock
Santosh had spent his career in banking operations, projects and enterprise businesses, including Microsoft. He wanted to work closer to the customer and the problem, with a smaller consultancy using AI to improve how work gets done. He approached the move from project manager to co-founder as a project of its own, with the familiar demands of time, budget and quality.
His question was how to jumpstart a forward-deployed engineering AI consultancy in four weeks. What belonged on the first to-do list? Which team and delivery method would make it work? Could it begin without a dedicated finance lead, and how would it sustain itself as the tools kept changing? Christian asked when the four weeks started. Santosh offered two possibilities: once the team and stack were ready, or once a customer had signed. The hour did not settle that clock.
Find someone willing to pay
Christian took Santosh's word jumpstart literally: a business needs somebody outside the founding team willing to put money behind it. His starting point was a customer. Brad made the test practical, describing ways to check willingness to pay before investing in a build, from a landing page and price tests to conversations with the people who would buy.
Santosh already knew people with processes they wanted AI to improve. They had problem statements, but did not know how to implement a solution. His difficulty was turning those conversations into an offer: what would he provide that other AI studios did not, and what could he charge without losing the customer or starving the team? Christian suggested learning from existing consultancies and testing pricing with people who might actually pay. No price was agreed.
Let the work size the team
Santosh wanted people and agents working together, with human responsibility for checking and deploying the work. Conor suggested thinking by business function, with a person in each role amplified by an agent team. AJ added that the size and scope of the customer's business would determine how much support was needed.
Santosh's own first estimate for a small customer was two full-stack engineers and a project manager. It was a sketch to discuss, rather than a staffing plan the room had tested. AJ described another way to approach the design: give agents the method used in a large transformation and ask them to carry as much of it as possible, while keeping people connected to the work and its stakeholders.
Become the customer's counterpart
Christian named a role that a team chart could miss. A client needs someone across the table who can develop the strategy with its leadership: where to begin, which work to improve, what capacity to free up and how the business will use it. That person has to understand both the business and the change it is contemplating. Brad put Santosh in that role.
The conversation had reached the question beneath the original list of tools and resources. Could Santosh build enough trust for a business to believe he understood it and could help it move? He agreed with that reframe. In his closing remarks, he returned to the value a customer would receive, the trust needed to earn an engagement, and the people still needed to deliver it.
Start where a team wants to change
AJ described looking for the bright spots inside an organisation: a team ready to improve one process or workflow. Start there, show what changes, and let the team's experience make the case for the next project. His example of turning a process from weeks into minutes illustrated that approach; it was not a result achieved for Santosh.
He also described walking the floor and listening to the people doing the work. A solution with their voices and fingerprints in it gives them a part in the change. Christian added a sequence worth keeping: understand the work, improve it, then automate it. The strategy and the motivation to follow it still needed human attention.
Deliver inside the client's constraints
On technology, Christian encouraged Santosh to stay open to the tools and keep learning. Brad brought the discussion back to delivery: a serious client may already have a tenant, a technology stack and compliance requirements that its consultants must use. Santosh recognised that constraint. The room offered a way to think about the stack, rather than selecting one for him.
Where the session landed
Brad's answer to sustaining the consultancy was to earn a first small win and build a continuing relationship from it. The retainer came up as a way to carry that relationship forward. Christian closed with the two purchasing criteria he had seen in consulting: competence and references. The first piece of work would have to earn its place as a reference; there was no client win to report from this session.
Several questions received direction rather than an answer. The hour did not choose between outside funding routes, resolve the finance role, set a final team size or work through a complete delivery playbook. It did not prove that a consultancy could be launched or a customer won in four weeks. The existing solution paper, prepared on 5 October, develops the conversation into a plan to test. It is an after-show synthesis, with additional material, and was not read back live.
Trying to turn your expertise into a business?
Tell us who you want to help, what they need and what is keeping you from the first engagement. We'll work the problem in the open.