The software consultancy model, reinvented
I've been thinking about how we sell software consulting, and something seems to be broken.
Throughout my career I've sold services for multiple seven figure projects, maybe surpassing eight figures total. The process was always the same. You make a connection. Someone knows you, or you know them, or pure cold outreach. Then you meet. And in that meeting, you sell capability. Not features, not deliverables. Capability. You show them you're genuinely excellent at what you do. You demonstrate intelligence. You walk them through your processes, show examples of prior work, explain the best practices that increase your surface of success.
You need to listen a lot. You need the client to feel heard, and you need to build confidence that you understand their challenge. They need to build trust that you're the one who understands it best.
The goal is simple: when hard times come, and they always do, your client needs to know they have an A-player on their team who will figure it out. Nobody buys services thinking "I want someone I'll have to constantly support and tell what to do."
Trust comes from knowing you're the best at what you do. Within a budget, sure, but the best within that budget.
But what happens when technology excellence changes completely? When the tools rewrite what "best" even means? AI tools, coding agents, knowledge bases. They've made building software look completely different. Excellence is no longer about recalling facts or knowing syntax. It's not even about showing that seven figure project you did last year, because the expectation now is that the same technology should cost five figures given the tools at your disposal. It may be the wrong expectation, but it's there. The Nifty IT Index dropped 20% in February 2026, the steepest monthly decline since 2008, with Jefferies warning that AI could shrink 22-45% of managed services revenue. Meanwhile, GenAI is the fastest-growing segment at TCS and Accenture, but it's cannibalizing legacy revenue rather than adding new business.
Picture this: a nimble company, five to ten people, all deep into Claude Code, Cursor, Codex. They have an internal knowledge base, agents generating documents and reports, AI analytics. They speedrun prototypes and MVPs. They get into a meeting and vibe code on the spot. They show you examples of how things could work, how things could play out, examples of inputs and outputs. They create datasets on the spot, show evaluation metrics. They use AI for security audits, test suggestions, Automated QA with computer-use AI. On the other side of the table sits the traditional consultancy. Built on a history of expensive software projects that delivered real impact, real excellence. But given current tools, not the most efficient way to build anymore.
The narrative is shifting. Questions about "Do you have Java experience on your team?" or "Do you have .NET developers?" will become trivial. Unimportant. Frankly, funny. The real questions will be:
- How do you use AI in your development process?
- What's your documentation process and how do agents fit in?
- What are your MCP integrations across software libraries?
- What's your security process for libraries suggested by AI agents?
- How do you manage architecture review, code review, and QA when you're looking less at every line of code and more at systems?
- How do you handle this higher level of abstraction?
Most companies are unprepared to answer these questions. The sales model of traditional consultancies breaks down.
I'm focusing on sales and management because engineering will adapt naturally. True engineers at heart will find the easiest and best way to solve technical challenges. The adoption will scale, with pains and hurdles, but it will happen. Companies will hire AI-native people who've seen how impactful these tools are. Engineering will be fine. But sales teams and management teams reinventing how they think about and manage software projects? That's the bottleneck over the next couple of years.
So how do you sell technical excellence in this landscape?
Get results quickly in front of users, in front of customers. Discovery sessions led by forward deployed engineers, by vibe coders, who by the end of the session have a proof of value ready. Not a proof of concept. A proof of value. I've been calling it that for years because you need to prove that by solving a particular challenge, greater value is unlocked. Think of it as a proof of concept but with real software deliverables, a real product at the end that your team and your customers can test, measure, and confirm the expected value is there.
Pair product managers who really understand the need being solved with hackers who can assemble a proof of value within hours. Validate it with customers on site. Show this new level of abstraction, this orchestration of software, right in front of potential buyers. Then over-deliver on architecture review, quality assurance, security, and scale.
The management process changes too. No billable hours. No tracking budget per feature. Instead: decision gates. Measuring impact. Following Northstar metrics. Unlocking budget as the metrics are reached.
The software consultancy model got reinvented. Most people just haven't noticed yet.