Thomas Bamforth and Pranav Sanghvi first met while both were participating in the MBA program at UNC’s Kenan-Flagler Business School. From the moment they met, Bamforth said, they knew they would one day be building something together.
That something turned out to be Carbonix AI—an emerging Chapel Hill-based startup that is aiming, on a broad scale, to accelerate the speed at which the life sciences industry operates. Specifically, Carbonix is using AI agents to enable Contract Research Organizations (CROs) to tackle proposals more swiftly and efficiently.
The idea for this startup was sparked when the pair met with an advisor who was speaking about his work at a CRO. During this conversation, Bamforth and Sanghvi learned about the long and outdated proposal process that CROs use.
When biopharmaceutical companies want to begin delivering new drugs, they need to set up clinical trials. Often, the companies will outsource these trials—and begin by reaching out to CROs to solicit proposals. This proposal stage is critical because it outlines the scope of the study, estimated timelines, costs, and other essential operational details.
What Bamforth and Sanghvi discovered in their discussion was that this process is still largely manual, slow, and fragmented across many CROs. Teams often spend 10 days preparing their proposals using outdated and manual operations.
“Once we sat down [and had conversations with CRO’s] and got a better understanding of all the different manual tests that go into this 10-day turnaround period, we started building up these tools that could really speed up and automate each of these processes,” Bamforth said.
Creating a multi-faceted solution
The Carbonix team aims to use their AI-powered tools to cut the turnaround times from 10 days to seven, five, even just one in the future.
The newest version of Carbonix AI is project-based, such that each Request for Proposal that a CRO receives has its own project view. Within this setup, there are various tools that speed up different parts of the proposal.
For example, the proposal process typically begins with a manual Request for Proposal (RFP) review. This often involves many hours spent on video calls wherein large internal teams from the CRO scrutinize the RFP line by line, trying to identify gaps or missing elements.
Instead of relying on these time-consuming discussions, companies can use Carbonix AI to develop a standardized, dynamic checklist tailored to their therapeutic areas and operational priorities. The RFP can then be automatically analyzed using AI—scanning the document against the checklist to flag missing components, ambiguous requirements, or misalignments with internal capabilities.
Another tool addresses budget comparison. A CRO may develop an extensive budget that could comprise thousands of spreadsheet lines, and then send that off to the biopharma company in question. The company will make extensive changes and then send that budget back.
“What we did was create an application that, for the most part, within 20 seconds can tell you exactly what changed,” Sanghvi said.
QUICK BITS
Startup: Carbonix AI
Co-Founders: Thomas Bamforth, Pranav Sanghvi
Founded: 2024
Team size: 2
Location: Chapel Hill
Website: www.carbonixai.com
Funding: Bootstrapped
Carbonix AI also speeds up the benchmarking process and the quality check. At a CRO, someone may be paid to comb through proposals and budgets (which, like a budget, may comprise thousands of lines on a spreadsheet), looking for errors with no true standardization. The process is simply based on the reviewer’s own experience.
“You can see why this is a really slow process, and you can imagine why there’s so much turnover with these organizations,” Bamforth said.
Carbonix currently uses large language models fine-tuned to specific use cases and works to stay up to date on which models are most effective for specific tasks. They have made their system very “plug-and-play,” and so are able to quickly implement new models based on updates.
Pilot projects and early success
Carbonix is currently underway with a pilot with a global, North Carolina-based CRO. The startup’s system is being used by both the company’s proposal teams in both the U.S. and Europe.
Bamforth noted that Carbonix’s strength lies in its specialized focus. Unlike larger companies, it can custom-build a platform specifically for CROs and CDMOs in life sciences and continually adapt it to meet user needs.
“Where we have seen success is being that custom tool for this industry, for this particular use case in the industry,” Bamforth said.
Carbonix will operate on a standard software-as-a-service model with a fixed fee per month based on expected usage and throughput.
In one year, Carbonix aims to be integrated into multiple CROs and CDMOs. Over the long term, they want to uplift the biotech market, and in particular the smaller biotechs trying to identify who the right vendors are for them.
