Deductions might be the most underestimated operational problem in consumer goods. When a retailer deducts from an invoice, for a late shipment, a promotional allowance, a compliance issue, the CPG brand gets a line item in a spreadsheet and a PDF that may or may not be accurate. Disputing it requires pulling data from multiple systems, cross-referencing against the original purchase order, and filing through retailer portals that were each built differently. Most brands just accept it.
Glimpse was founded in 2024 by Akash Raju, Anuj Mehta, and Kushal Negi, who met at Purdue. They pivoted hard from an earlier idea and found this one. Within a year they had 10x revenue growth, 150 brands on the platform, and a $35M Series A led by Andreessen Horowitz, bringing total raised to $52M. That is a fast-moving team.
The Market
A quiet problem that sits across every CPG company's books.
Retailers take 7 to 15% of gross sales in deductions. For a $10M brand, that could mean up to $200,000 lost per year to invalid deductions alone. The trade promotion management market sits at roughly $1B in 2026, projected to reach $2.2B by 2035. But that number understates the actual addressable problem: Glimpse is going after the broader retail operations infrastructure layer, not just the deductions slice.
The scale of what is still manual is striking. A Glimpse AI agent reviewed 17,000 deductions for a $1B CPG company in under 24 hours. That same work done manually would take nearly two years. Most of the industry is still doing it manually.
The Landscape
Where the existing tools fall short.
automation
normalization
coverage
resolution
● Purpose-built◐ Partial / legacy○ Not a focus
HighRadius is the established player in accounts receivable automation, but it is a broad platform built for large enterprises and was not designed for CPG-specific deduction workflows. iNymbus and SupplyPike use RPA approaches, which automate repetitive steps but break on unstructured inputs. Glimpse's AI agents operate on unstructured data natively, which is a different claim about what the system can actually do.
The Engineering
The normalization problem comes before the AI problem.
The engineering challenge Glimpse is taking on is not just automation. It is normalization across dozens of retailer formats and data schemas that were never designed to talk to each other. EDI files, PDF invoices, proprietary portal exports from Walmart, Target, Amazon, KeHE, UNFI. Each built differently, each requiring a different parsing approach. Getting the data pipeline clean enough for a model to be useful is the hard part that has nothing to do with model quality.
I spend most of my time on exactly this kind of problem. At Movius, I built pipelines that archive carrier metadata across inconsistent formats into object storage. At Amazon, I built a tool that reconciled pricing data across S3 and DynamoDB. The unglamorous data normalization work always turns out to be the hard part. The teams that solve it build durable things.
The hard part is not the automation. It is the normalization that has to happen before any model touches the data.
The Frontier
From deductions tool to CPG operating system.
Glimpse's latest announcement frames the company as building AI-native infrastructure for CPG and retail broadly, not just a deductions product. That is a meaningful shift in ambition. If the normalization layer is working well across 150 brands and multiple retailers, the platform becomes the integration layer for everything that touches retail cashflow: cash application, trade planning, promotional reconciliation, inventory disputes.
The a16z bet reads that way. A $35M Series A led by Andreessen Horowitz for a company that is less than two years old, 10x revenue in one year, and a hard pivot that produced product-market fit faster than most companies find it the first time. That is a team worth paying attention to.