Longeye Raises $20M Series A for Investigative AI
A detective following an AI-generated finding needs a quick way back to the thing that actually happened. Longeye puts that return trip inside its investigative workspace, linking findings to the documents, messages and recordings behind them. The commercial promise depends on making review practical when a case contains more material than a person can comfortably examine.
The San Francisco company announced a $20M Series A on October 21, 2026, led by N47 with participation from Andreessen Horowitz’s American Dynamism practice. Longeye says more than 50 justice agencies have used the platform across more than 2,300 investigations. The funding supports broader adoption and an expanded ambassador team that helps agencies introduce AI into investigative work.
That combination gives the transaction its significance. Longeye is financing both the software that finds a useful passage and the human work required to make checking that passage a routine part of an investigation. For public-sector AI suppliers, the training conversation is becoming part of what the customer buys.
N47 backs founders who know the deployment problem
Cofounder and CEO Guillaume Delépine previously built the public-safety business at Skydio. Cofounder and CTO Danielle Man led platform engineering at Apollo GraphQL, according to Longeye’s founder biographies. Their experience brings two relevant disciplines into the same company: getting technology into agencies and building the infrastructure underneath a product that people rely on.
N47’s investment thesis, published by T.J. Rylander and Gautham Pasupathy, describes conversations with customers and identifies both founders as alumni of its portfolio. That history gives the investment a specific institutional connection. N47 has experience with the environments from which the founders emerged, alongside an interest in the evidence workflows they are now addressing.
The Series A follows a previously disclosed $5M seed led by a16z American Dynamism, with Seven Stars Capital and strategic angels participating. The earlier financing belongs to a different round; the October announcement names N47 and a16z for the Series A. Together, the disclosed rounds account for $25M, while the new transaction contributes $20M. The announcement does not supply a valuation.
The product puts a source behind the answer
Longeye’s digital evidence tools work across video, audio, phone extractions, financial records, documents and social-media material. Investigators can ask questions in ordinary language, build timelines and examine relationships across different evidence types. A result is intended to lead the user to the relevant source rather than require another search through the entire collection.
That changes the economics of attention inside the case. A recording can contain the passage an investigator needs while still demanding hours to locate it. Connecting a finding to a timestamp makes the next act of human judgment smaller and more concrete: listen to this passage, examine its surroundings and decide what it supports. The product’s usefulness is tied to the quality of that handoff.
A link also makes an answer easier to challenge. A cited statement can be misunderstood, stripped of context or assigned too much significance, so the presence of a citation leaves an important job for the person reading it. The practical purchasing question becomes how well the system helps an investigator inspect the underlying material and revise an initial interpretation. Faster retrieval creates value when that inspection can fit into an actual working day.
Adoption includes the people who teach the pause
The company’s responsibility page describes a report workflow in which investigators accept or reject individual sections, original evidence remains unchanged, and the author provides an attestation before export. These are vendor-described product controls. Their design puts an identifiable person between generated prose and a report that leaves the workspace.
The Series A also supports expansion of the Ambassador Team, led by Denton Carlson, the retired San Ramon police chief. The group helps agencies with implementation, training and responsible adoption. Experienced practitioners can connect a product feature to the circumstances in which a detective will use it, including the crowded shift when a review step is easiest to hurry past.
For Longeye, that makes the cost of deployment broader than creating an account and uploading files. Agency leaders need their teams to understand what an output represents, how to check it and when to seek more evidence. An ambassador can help translate those expectations into habits. The company is committing resources to that translation as it expands beyond its existing agency base, keeping the adoption effort close to the professional practice the software enters.
Evidence access reaches both sides of a case
Longeye also serves eligible public defender offices and defense nonprofits through a grant program. Its public-defense offering describes source-linked review of discovery materials, giving a defense team a way to examine the records it receives. The same collection can raise different questions depending on who is responsible for reading it.
That breadth adds a demanding dimension to the company’s expansion. A tool used by investigators and defenders has to remain useful to people whose interpretations may diverge. Access controls and source inspection sit close to the core service: each office needs to examine its own material without losing control of it. Longeye’s security documentation describes case-level permissions, audit trails, US data residency and restrictions on using evidence to train language models. Agencies can request the underlying security materials for their own evaluation.
The reported agency and investigation counts establish the company’s account of deployment, while independent accuracy measures and financial performance remain undisclosed in the reviewed announcement. As Longeye grows, another part of the service will be repeated office by office: helping the person reviewing a call, a spreadsheet or a draft report return to the source and decide what belongs in the case. The ambassadors are being funded to help that work survive the trip from a product demonstration to an investigator’s desk.
AI & Machine Learning funding, last 30 days
DevCuration's funding database tracked 25 AI & Machine Learning rounds totaling $1.4B in disclosed capital over the past 30 days. Recent deals we covered:
- Azoma Raises $11.2M Series A for AI-Shopping AttributionSeries A · $11.2M · Oct 22
- Vectorial Raises $3M Seed for Human Behavior SimulationSeed · $3M · Oct 21
- AuxoAI Raises $20.7M Series A for Forward-Deployed AISeries A · $20.7M · Oct 21
- Different Raises $48M Seed for AI Demand GenerationSeed · $48M · Oct 20
- Tolan Raises $30M to Build a More Personal AI Companion$30M · Oct 20
Frequently Asked Questions
Why does source-linked AI matter in an investigation?
A source link lets an investigator return to the document, message or recording behind an AI-generated finding. It supports human inspection of the result; the investigator still needs to evaluate context and meaning.
How does Longeye’s Ambassador Team fit the funding strategy?
The Series A supports expanded agency adoption and the Ambassador Team. Experienced practitioners help agencies introduce the software, train investigators and develop habits around checking AI-assisted findings.
How is the Series A different from Longeye’s earlier seed round?
The October 21, 2026 announcement is a $20M Series A led by N47 with a16z American Dynamism participating. The previously disclosed $5M seed was led by a16z and included Seven Stars Capital and strategic angels. Those disclosed rounds account for $25M in aggregate.
Can public defenders use Longeye?
Longeye says it serves eligible public defender offices and defense nonprofits through a grant program. Offices should confirm current eligibility and availability with the company, because the website also retains an earlier priority deadline.
What do Longeye’s deployment figures tell buyers?
The company reports use by more than 50 justice agencies across more than 2,300 investigations. Those figures describe deployment and workload; they do not by themselves establish model accuracy, financial performance or case outcomes.
Where the Money Moved
The intelligence briefing of the innovation economy. Funding, M&A, debt and fund closes, read as market signal rather than deal announcements.
Subscribe to Where the Money Moved