Picture this: a potential new client calls a small firm at 18h30 one evening. They have a quick question regarding immigration law and that quick question is likely to turn into a brief for the firm. Only problem is, the receptionist has gone home and nobody else hears the phone ring. By 08h30 the next morning, the potential client has reached one of the other firms recommended to them.
A simple missed call, but still, a missed opportunity. Should that happen multiple times a month, the impact compounds.
Paravo, a legal tech player that emerged from stealth on 6 August 2026, is built around a specific observation about that phone call and the 70% of a small firm’s day it stands in for. In the words of Cesar Tapia, the long and short of the founders’ argument is that flat-fee firms in areas like immigration, personal injury, family law and employment spend roughly 70% of their time on non-billable work: chasing leads, handling intake, managing calendars, following up with clients, updating the CRM, sending messages, and everything happening around the legal work rather than the legal work itself. The remaining 30% is what the firm gets paid for.
Paravo aims an AI-native platform at that 70%. It handles lead generation, responds to inbound enquiries the moment they arrive (message first, with voice AI in development), runs an AI intake receptionist that answers phones 24 hours a day and books consultations directly into a firm’s calendar, and later runs reactivation campaigns to previous clients whose next legal need is predictable from the last matter. The four modules inside the product are Leads, Intake, Outreach and Insights, and the company describes itself as a ‘revenue engine’ rather than an intake tool.
This week, The Legal Wire sat down with Cesar, who co-founded Paravo with Eslam Odeh. Cesar spent his early career on the investor side, working in venture capital across the UK and Spain for around three years before joining Carwow, the UK car-buying marketplace, where he met Eslam. The Paravo idea came out of their own experience as immigrants navigating visa applications, and their decision to ask their own immigration lawyers what the real problems in running the business looked like.
Why Paravo has decided not to build another Harvey
Cesar explains their approach in a bold way: it is a deliberate refusal to build the product the wider legal AI market would have expected. Enormous amounts of venture capital have gone into all-encompassing legal AI platforms aimed at the substantive legal work: Harvey, Legora, and a growing list of others raising rounds at valuations that require them to eventually replace significant portions of lawyers’ substantive daily practice.
Cesar’s view is that the substantive-legal-AI space is crowded, and the capital being raised into it creates expectations most of those companies will find difficult to meet. “I’ve seen many companies build genuinely successful businesses with happy customers,” he said, “but because they raised so much money at such a high valuation, the expectations became enormous. Eventually, they needed to raise even more money.” His argument is not that any specific legal tech company will fail, but that the financing structure attached to their category is a separate problem from the product, which founders in that lane will spend increasing energy managing rather than building.
Paravo’s move is to look somewhere else entirely. Not at the substantive legal work the lawyer does, but at what surrounds it. The firm’s bottleneck, according to Cesar, is the 70% of the day that is not substantive legal work. And the market most affected by that 70%, the flat-fee practices, is fragmented, competitive, and under-served by the legal AI conversation.
The flat-fee firm is a different kind of business from the Big Law model in that the incentive is not to bill more hours but to close matters more efficiently, because the fee has already been agreed. That gives the firm a structural interest in speed a billable-hour firm does not have, and gives Paravo a customer whose incentives are aligned with the product’s value proposition. In a billable-hour firm, the maths does not work the same way.
TLW: You’ve deliberately gone against the venture money by focusing on the non-billable seventy per cent and targeting flat-fee firms. What is the rest of the legal AI market missing here, and where could Paravo get the bet wrong?
Cesar: “The market is chasing the wrong problem. Everyone builds better document review tools… that’s a commodity race.
I don’t think the Harveys/Legoras are competitors because they’re building a different product for a different customer. Harvey and Legora are training into hourly-bill firms, and they sell to investors that they have the data, and they will eventually replace the firm to sustain the valuations they are getting.
Where are we wrong? Customisation cost can scale faster than revenue. We have learned from Palantir’s Forward Deployed Model, which has worked very well for the first clients to ensure we shape the product based on firms’ requirements, but we believe building a tool that does not need weeks of customisation is key to having good unit economics.”
What the product does, and the small detail that gives it away
The product has four connected pieces, arranged around a simple thesis: a lead is worth the most in the first minute after it arrives, and rapidly loses value from there.
The Leads module handles client generation, through Paravo’s advertising and lead-marketplace relationships or the firm’s existing sources. The Intake module handles inbound enquiries and phone calls, with an AI receptionist that screens the caller, gathers the details the firm needs, and either transfers the call or books a consultation directly. The Outreach module handles reactivation, tracking previous clients and predicting when their next legal need is likely to arrive (an expiring visa, a family event, a statute date) and sending personalised outreach at the right moment. The Insights module is the analytics layer.
The detail that gives the product away is in the intake demonstration on Paravo’s website. A caller phones in, says they have been in the country for 10 years and have not applied for a green card. The AI receptionist explains that an attorney would need to review the situation, offers an hour online consultation for USD 250 credited toward legal fees if they proceed, checks tomorrow’s availability, and books 10h00. The booking already contains a specific price, a specific credit-toward-fees structure, and a specific set of qualifying questions a real firm would want asked before the attorney’s time is committed.
In this regard, the platform is more than a voice model. It’s moving into the territory of selling the business logic a small firm would otherwise have to think through, train someone to deliver, and constantly correct.
When asked what makes the platform useful in practice, Cesar mentions customisation. “We have a product, but more importantly, we’re building a solution,” he said. Depending on the firm, Paravo activates the modules that fit and integrates them with the firm’s existing CRM (HubSpot, Lawmatics, LEAP, and others), calendar (Google, Outlook), and case management system (Clio, with Zapier integrations for the rest). Some tax firms, for example, receive few inbound calls but need help with reactivation. An immigration firm may have the opposite pattern. The value, Cesar argues, is in configuring the engine around the firm’s specific needs.
TLW: Customisation-per-firm is expensive to deliver, and most SaaS companies specifically try to avoid it because it does not scale. How do you keep it genuine as you grow beyond founder-led sales, and where does it become a constraint rather than an advantage?
Cesar: “Customisation only scales if you build it into the architecture, not bolt it on after. We’re designing for variation from day one. Different firms activate different modules and different configurations based on their intake patterns and staff setup.
Customisation is expensive, but it’s more expensive to acquire a client that churns because the product does not adapt to their needs.
We believe the current risk isn’t being too customised, as everything we build for each client makes the product stronger and each new client takes less time than the previous one to be up and running.”
On selling AI to lawyers who do not want to hear about AI
A topic that always interests me is the peculiar sales problem legal AI companies face, which is that a meaningful share of their prospective customers actively do not want to be sold AI, and it’s not that the customers are wrong, per se, but that “AI” has become a loaded word for many small firm owners.
It’s for this reason that Paravo chooses not to use the word AI at the start of the conversation at all. They show the product first, walk through what it does, and let the firm react. “Their reaction is often, ‘Wow’ and next, ‘how does this work?'” he said. “Then we explain the technology behind it.” There are also firms that seek Paravo out precisely because they want AI. Both routes end in the same conversation.
The related question of AI disclosure to a firm’s own clients, Cesar leaves with the firm. Paravo’s own view is that end clients want their problem solved and are focused on the outcome. But if a firm believes its clients should be told they are speaking with AI, Paravo builds that into the configuration. What this tells you is open for debate, but it seems that the founders are not trying to win an argument about whether disclosure should be required. To me, it leans towards demonstrating that the founders are trying to build a product that works whichever way the firm has decided the argument.
TLW: What single behaviour separates the firms that adopt AI successfully from the firms that buy it and never use it, and what should vendors be doing differently for the second group?
Cesar: “You need a champion. Someone inside the firm who wants to make a real difference and is willing to drive change. The firms that succeed have a partner or office manager who sees the problem, commits to fixing the workflow, and pushes the team through the transition. Failed adoptions happen when you sell to a partner who nods along but doesn’t actually care about changing how things work.”
What the market is misunderstanding about small firms
The observation Cesar mentions more than once is that small law firms are running businesses of real financial substance on remarkably basic technology. “You can have a relatively small firm with 5 – 10 people generating significant revenue, yet they’re still managing important parts of the business through Excel spreadsheets and other very manual processes,” he said. A firm turning over meaningful money and running its client acquisition on a spreadsheet is one competitor away from losing a significant share of its future work to whoever gets to the leads first.
Cesar’s view of the next three years is that people are unlikely to keep manually updating CRMs in the way they do today, and if they are doing it manually, it will be because they specifically want to. That inverts the usual question. The question is not “should I adopt AI?” It is: three years from now, when the technology is available and cheap and my competitors are using it, what is my defence for continuing to do this work manually?
There’s also a related point about market dynamics. Flat-fee immigration and family law firms do not have the client-relationship moats Big Law does. They compete on service, speed, and price. That makes them difficult customers in the short term (cost-conscious, slower to adopt) and interesting ones in the medium term: once one firm in a local market pulls ahead, the pressure on the others is immediate.
Whether Paravo becomes the platform those firms adopt, or whether that role is played by one of the intake competitors already in the space, is what will likely emerge in the next 12 – 18 months.
TLW: If you were advising a managing partner at a five-to-ten-person flat-fee firm, what is the first specific investment you would tell them to make, and what is the mistake most firms that size make when they decide to adopt technology?
Cesar: “Don’t change your entire stack. That’s the biggest mistake. Firms get excited and adopt five new tools at once. They blow up their workflow chasing the perfect tech solution.
I would suggest they work with someone like us to automate what they already have… keep doing what currently works, just with less manual work: Get the receptionist off the phone. Get the partner out of the CRM. Then iterate. Once you’ve eliminated the chaos in your current setup, you’ll see what actually needs to change next.”
Paravo in the bigger picture
Paravo is a new company, launched a few weeks ago, with paying customers, a conviction about which part of the market is worth building for, and a founder that’s pretty candid about what he thinks the rest of the legal AI conversation is not getting right. While the company has done a pre-seed raise to build its first product, Cesar notes that its nowhere near what the ‘big names’ are raising. His view is that the financing structure attached to the loudest legal AI companies is a problem those companies will have to manage in ways that shape their product decisions. Paravo has chosen not to have that problem yet.
Whether the revenue-engine framing turns out to be genuinely different from the intake-plus-outreach tools already in the market, or an unusually well-executed version of a category that will eventually consolidate, is what the next year will show.
What is worth noting now is that Paravo has taken a specific position on a problem the rest of the market has been happy to leave under-covered, has committed to a customer segment the largest legal AI companies are not particularly interested in, and has a founder whose account of why the industry has ended up where it is worth reading. For a new kid on the block, that is a considered start.
