Hello,
In late 2025, I started Pravah Capital to acquire and operate a single Indian B2B services or technology business. 90 days in, three things are already clear: the deal flow is real, the succession gap is real, and the valuation gap is real. We evaluated 64 situations this quarter, are actively pursuing nine, and have submitted a non-binding offer on one.
Many of you, in some way, played a part in getting me here. What follows is the texture of what 90 days inside this market actually looks like - the first of what I hope will be a quarterly note from the field.
The succession gap thesis is real, and the path is naturally a patient one. A generation of founders in their 50s and 60s with no obvious next chapter is as deep as the data suggested. The journey from a first conversation to a transaction takes time, which makes sense. The bigger thing being built is trust. A founder handing over a company they have spent thirty years on does so on much more than price.
My working assumption is that the right buyer relationship typically gets built over six to twelve months, and that the moment a founder is genuinely ready to transact usually comes after they have spent some time in the market. I want to be the first call when that moment arrives. This is a search that will reward patience and the relationships we build over the next year, and I am optimistic about the shape of what is in front of us.
The Indian SME advisor ecosystem is far more developed than I had expected. A fragmented layer of four-or-five-person boutiques, deeply embedded in local founder networks, is now the primary engine of deal flow. Contrary to what India's M&A landscape looked like even five or ten years ago, this intermediary base has genuinely ramped up in recent years. The implication: a two-person team can evaluate real volume if plugged into this network. Q2 priority is shifting toward proprietary sourcing, but this layer is what makes the current velocity possible.
AI is reshaping both what we look for and how we operate. On the investing side, every target now has to clear an AI-resilience test before we deepen: businesses where AI augments rather than substitutes, where domain expertise is hard to encode, where pricing is on outcomes rather than headcount, where embeddedness in customer workflow is genuine. This filter has killed several otherwise interesting situations. On the operations side, my associate Manav and I are running the search through a custom Claude Code stack that lets two people cover the ground a five-person team would have covered a year ago. Sector primers, deal screens, advisor management, contact enrichment - most of what used to be analyst-week work compresses into hours. The composition of the team I will eventually need looks different from what I planned twelve months ago: less routine research, more judgment and relationship work.
My working belief: this is a market where patience and relationship depth will matter more than speed, and where disciplined buyers will ultimately have the advantage as valuation expectations reset.
What follows is more on what we are seeing and what I am wrestling with. If you have ten minutes, the second section is the most useful. If you have twenty, scroll all the way down - I have two real questions I would love your read on.
Bhavik
Sixty-four deals across roughly seven sector clusters. Below: six anonymised situations from the quarter that capture the kind of decisions we are navigating, then the filters we are applying across sectors.
Anonymised, but real - each is a live or recently-evaluated deal.
A business at ~$14M revenue running mission-critical infrastructure for several hundred Indian stockbrokers. SEBI-mandated upgrade cycle is creating a forced refresh. Strong stickiness, high switching costs - but the seller is anchored to peak-cycle multiples and one year of decelerating growth has not yet moved that anchor.
Our bet: integrated infra and broker SaaS benefit from rising market scale, complexity, and regulation, with durability driven by switching costs.
A dealer-management platform at ~$8M revenue embedded in the workflows of large OEMs across auto, FMCG, and pharma, with a 25-year track record and 85%+ recurring revenue. They are voluntarily winding down legacy one-off revenue to reinvest in AI products, which compresses growth and margin in the near term. The underwriting question becomes: do you pay for the bounce-back, or wait for proof?
Our bet: incumbency and a 25-year data moat sustain value despite weak new-logo growth.
A pan-India equipment-handling operator at ~$4M revenue with 49% EBITDA from a non-obvious source: disciplined sourcing of secondhand machinery. The business is more capital-intensive than we typically pursue. The question is whether the sourcing edge is a real moat or just a moment.
Our bet: service density and uptime SLAs are the moat, not asset ownership.
A chip-design house at ~$7M revenue operating at the most advanced process nodes. AI augments rather than substitutes the work. Globally scarce skill set, deeply embedded in foundry workflows. We have submitted a non-binding offer here. The valuation conversation is open.
Our bet: at the frontier nodes, AI accelerates the rare expert rather than substituting them - a position too costly to dislodge.
A back-office services business at ~$5M revenue serving the US mortgage-default market. Strong growth, 80%+ repeat revenue, structural cost arbitrage. But 65-80% of the underlying work falls into among the most automatable categories in the AI mapping we use. The deal lives or dies on whether the service has a credible pivot path.
Our bet: compliance-critical mortgage servicing remains human-in-the-loop despite AI, preserving margins.
A regulatory-driven testing lab at ~$3M revenue with 50%+ EBITDA - well above what our asset-light orthodoxy says should be sustainable. Accreditation moat, several thousand active clients, compliance-driven demand. The intellectual question is how much of the margin is structural versus cyclical, and whether scale can be replicated.
Our bet: accreditation and consolidation drive platform outcomes in a fragmented testing market.
Across all of the above, the principles we filter for are consistent.
Manav Chaudhry, who joined as our first full-time associate this quarter, is the other half of the team. Everything runs through a custom AI-augmented stack we have built to integrate sourcing, screening, sector research, and outreach. Every deal evaluated, sector primer built, and advisor interaction compounds into the next. What looks like overhead in Q1 is, I expect, a structural advantage by Q4 - particularly in decision speed and pattern recognition.
Two questions I would love your read on. Both are live; both have shaped real decisions in the last 90 days.
My working heuristic is to favour businesses where AI augments rather than replaces, where domain expertise is hard to encode, where pricing is on outcomes rather than headcount, and where the service is genuinely embedded in customer workflow or data. This filter is doing real work - it has killed deals - but it feels under-tested.
The honest version of the question: which of these signals would still hold up if the next two years of AI capability moved much faster than expected? And which businesses look protected today only because we have not yet seen the right disruption vector? If you have seen this play out, either in investing or operating, even a short note on how you have approached it would help.
Our framework says: asset-light, recurring or repeat revenue, deeply embedded in customer workflows, sector limited to B2B services and tech. It is a useful filter - it kills most of what we see. But quarter one has put real exceptions in front of us: a capital-intensive industrial services business with 49% EBITDA from a non-obvious sourcing edge, and a regulated business with margins structurally higher than our framework expects.
The question I would value your read on: when a real exception lands on your desk, how do you stress-test whether it is genuinely exceptional, or just a story you are telling yourself to break a discipline you should have kept? Particularly interested in pattern recognition from situations where you chose to break your framework, and how that played out.
I'll send these every quarter or so to keep you updated. If you would rather not be on the list, just reply and I'll take you off, no offence taken. Equally, if anyone in your world would find this useful, feel free to forward.
If you have a strong view on any of the above, I would value the perspective. And if you are in Mumbai, would enjoy continuing the conversation in person.
Thank you for the early support.
Bhavik