AI agents are missing all the discussions your team is having. SageOX has an answer: agentic context infrastructure

As AI model providers are increasingly moving downstream, launching products and agents for specific enterprise applications and areas such as finance, a big question still remains: How will said AI agents be equipped with the appropriate features? Context Scoring a task – who assigned it, which other stakeholders are involved, what data or discussion has occurred about it and how should it be done?

this practice of "reference engineering" AI remains one of the great unsolved problems of the era. But SageOx, a Seattle-based startup founded by veterans who built the original AWS EC2 and EBS infrastructure, believes there is an answer: a new system layer it calls "Agentic reference infrastructure."

Using a combination of small hardware recording devices and existing applications enterprises already rely on – Slack, email, documents, files – and applying new, open-source frameworks and instructions on top of it all, SageOX has developed a system by which enterprises can hire agents "in the loop" And the functions of the enterprise are updated like their human employees, and they are prevented from doing so "flowing" Away from their assigned tasks and larger goals of the firm.

“We’re capturing this entire context where this occurs," Ajit Banerjee, founder and CEO of SageOX and former Hugging Face, Meta, Amazon and Apple engineer, said in a recent video call interview with VentureBeat. "Product development is a team sport, and context doesn’t just come from people typing on keyboards. It happens in conversation.”

by occupying "Why" Back "What"- The intention that lives in Slack threads, whiteboarding sessions, and water-cooler conversations – SageOx aims to provide a "Hive Mind" This ensures that agents do not drift and humans remain in flow.

"The way people work is not old-school coordination, where I write an issue and then it goes through a sequence. It must be almost like playing jazz," Banerjee added.

Today, the company quietly emerged to announce a $15 million seed round led by Canaan and participation from A.Capital, Pioneer Square Labs, and Founders Co-op.

Team Memory Architecture

Today’s AI agents work in isolated sessions, lacking a shared memory of prior decisions or architectural intent.

Each task effectively starts from scratch, forcing developers to manually recreate the context – a process that undermines the speed that agents provide. SageOx addresses this through a multi-surface product suite designed to capture context wherever it naturally occurs.

is at the center of this ecosystem ox dot. A customized hardware device designed for the shared office, Dot captures meetings, standups, and design reviews with a single touch.

Its biggest feature is "auto rewind"-A fail-safe for a team’s innate talent. If there is a breakthrough during an unrecorded conversation, auto rewind allows the team to "go back" And capture the discussion after the fact. This audio is transcribed, identified by the speaker, and distilled into the team memory, where it becomes accessible to both humans and agents.

For developers, open-source, MIT-licensed bull cli Provides a bridge. such orders ox agent prime Allow coding assistants to consult the team’s shared history before writing code, including Cloud Code and Codex. This ensures that if a team has decided to use a specific authentication pattern in the meeting, the agent knows without having to explicitly state it in the prompt.

As Dr. Rupak Majumdar, Scientific Director of the Max Planck Institute for Software Systems, said after seeing the team’s development pace, they are effectively "Treating code like assembler."

Agentic Engineering: Moving Beyond "Clean" code

The shift to agent-first workflows has forced the SageOx team to rethink nearly every principle of modern software management.

SageOX CTO Ryan Snodgrass, formerly of Amazon, notes in a blog post transcript that traditional branch management and "Clean" Commit histories are often "Bad for agents." In the old world, humans preferred larger PRs that were easier to read during a single code review.

In the agentic era, 10,000-line PRs spread across the codebase make it impossible for an agent to reason about intent.

Instead, SageOx advocates small, high-volume, and highly focused commitments. it "agent-readable" History allows the machine to look back and understand why a specific change was made. The team is also reevaluating repo structures; While they currently use a monorepo for 750,000 lines of code, they are exploring a future where agents can manage a group of micro-repos, as agents can "Go to hell" When a codebase becomes too large for their context window.

this philosophy "speed-over-stasis" Despite having no recent hardware experience, the team was allowed to create their own firmware for the Ox Dot in less than two weeks.

By feeding technical PDFs and documentation into the AI ​​model, they bypassed months of traditional research. CEO Ajit Banerjee says this "unlearning" of old habits—realizing that "uncombined heavy lifting" Knowledge work can now be uploaded into a system that remembers everything the team knows.

Radical Transparency: Beyond Open Source "open work" Sample

Perhaps just as important is SageOx’s commitment to technology "Open work." Moving beyond traditional open-source software, the company is practicing radical transparency in an effort to boost the pace of development across the entire open-source community and any enterprise that wishes to learn from the way they work.

The SageOx team openly shares their internal signals, their planning sessions, and even their unfiltered internal debates with the public. Users can sign in to the SageOx console and watch the team build SageOx in real time.

it "open kimono" The approach was a deliberate decision to lead by example. Banerjee argues that since they are asking teams to change the way they work, they should be prepared to show that "WTF" Moments and course corrections as they happen.

"The revolution will not be televised," Banerjee says. "It’s going to be SageOxed."

The purpose of this transparency is to prove that a small, lean team-"yoking lean"-Can outperform larger organizations by leveraging a shared context layer.

On how SageOx plans to monetize and become profitable, Banerjee said the revenue path is based on the AWS EC2 playbook: Start with early adopters, especially small AI-native startups, then expand toward enterprises as the need becomes clear.

Genealogy of infrastructure

The technical foundations of SageOx lie in the early days of cloud infrastructure.

Banerjee was an original member of the AWS EC2 team, and Snodgrass was one of the first engineers at Amazon to lead the transition from monolithic architectures to microservices.

This background is reflected in the company’s name: The "Bull" represents "yeoman work" Their goal is to make that happen – a reliable beast that can handle heavy loads of data and context so the team can move forward.

The SageOx vision is one where humans are no longer manual assemblers of context.

Instead, they act as directors of a "parallel processing" engine.

In a recent demonstration, a feature request went from verbal discussion to full implementation within seven minutes. By priming the coding agents with the recorded context of the original discussion, the team bypassed the need for formal specifications or Jira tickets.

new way of working

SageOx is currently focusing its efforts on "ai-native" Startup-teams that work primarily through signals and rely heavily on agent colleagues.

Their suite of tools, from the open-source Aux CLI to hardware-enabled Aux Dot, is designed to solve the immediate problem of alignment drift.

As AI moves from a tool to a teammate, the most valuable asset a company has is no longer its raw source code, but its shared context.

SageOx suggests that the way forward is not to hold back information. "private fence," Rather, the aim is to create a communal foundation where the intentions of every teammate – human or machine – are visible. In this new era, the teams that win will be those that are able to remember as quickly as they can execute.

<a href