An analyst at an investment bank can spend a week on a pitch book that a managing director reads in twenty minutes. Comparable company tables, precedent transactions, market data pulled from three different systems and checked twice: most of it is assembly rather than judgement.
Mpho Montsho knows that work first-hand. Before co-founding Techbridge Innovations, a US-based software development house, he spent two years as an investment banking analyst at Deutsche Bank, working on transactions including the initial public offering of digital asset platform Bullish and the sale of insurtech company Vouch to Hiscox.
"A lot of what junior bankers do is make sure nothing is wrong," he says. "Every number in the deck has to tie back to a source. That's important work, but it's also exactly the kind of thing an agent can do well, as long as someone senior is still checking the conclusions."
At Techbridge, Montsho leads product, operations and the delivery teams, and much of the company's work is for investment and asset management firms. He sees banks and asset managers arriving at AI agents from opposite ends of the same problem. Banks produce enormous amounts of analysis for clients and deals. Asset managers consume it, along with filings, earnings calls and their own research, and have to decide what it means for their portfolios.
.jpeg)
Where agents are already working
On the asset management side, Montsho was product manager on an AI analyst agent for a client's portfolio management team. He defined the agent's scope with the portfolio managers and guided the engineering team through delivery across three milestones. It connects to the firm's research, email and portfolio data, and the portfolio managers use it daily to check what has changed on their holdings and whether new information challenges their investment views. The second phase, which he is now managing, adds alerts when an investment thesis appears to be drifting, tracking of conviction across email threads, and quantitative analysis inside the chat.
He was also product manager on a platform for buying and selling private businesses, which brings business owners and acquirers into one system with AI-assisted analysis of the deals coming through. It is work that draws directly on his time in a bank deal team.
What changes, and what doesn't
In banking, Montsho expects agents to take on more of the preparation: first drafts of comparables, screening of potential buyers or targets, and checking documents in due diligence. In asset management, he sees firms building bespoke agents around their own process, holdings and house views, which help portfolio managers weigh new evidence and decide how a portfolio should be composed.
In both cases, he says, the decision stays with people. "A managing director isn't going to sign off on a valuation because an agent said so, and a portfolio manager isn't going to buy a stock because of a chat answer," he says. "The agent's job is to get them to the decision faster, with the evidence in front of them."
He also thinks generic tools will only go so far. "Every bank and every fund has its own way of working. The agents that get used are the ones built around that."
Montsho, who lives in Johannesburg, is currently expanding Techbridge's work with investment firms alongside the second phase of the portfolio management agent.
