Backed by Pareto Holdings, Cory Levy, Z Fellows
The Human API
for the enterprise employee
Get Early Access
The Human API We turn human expertise into queryable endpoints, collapsing the management hierarchy so teammates and AI agents have direct access to context without ever waiting for a person to be available.
Scroll to explore ↓

Human availability is the speed limit
of every company.

AI is getting faster. Your team isn't. The corporate hierarchy turns your best builders into human databases — the only path to why that decision was made, where that config lives, who owns that service. When that person is asleep, in a meeting, or deep in flow — everyone downstream stops. Coordination already consumes 57%* of the average workday, because we are still routing knowledge through human-speed chains of command.

*Microsoft Work Trend Index, 2024

We believe the next era of AI-native teams requires an intelligent coordination system between people that allows us to…

01
Learn
Build a living map of artifacts and owners
Who owns what. Who knows what. Who's blocked on whom. A self-updating network of Human APIs that indexes not just documents, but how work actually flows across your team.
02
Orchestrate
Automate 90% of routine friction
90% of what middle management routes — status updates, context handoffs, "who do I ask" — flows automatically. Employees stop being information bottlenecks and become decision-makers.
03
Compound
Turn coordination into organizational memory
Every answer, every handoff, every resolved question compounds into organizational memory. The 100th hire inherits the context of the first — without a single onboarding meeting.
How it works
Connect
Understand
Observability
Answer
Interface
01
Connect
#
Slack
RK
"How are our relations with Oscorp?"
M
Meeting
JC
"We decided to go with DynamoDB"
PR
Code Review
LW
"This breaks the auth flow"
RK
NP
JC
LW
TH
AV
DMs: never
1-on-1s: never
HR: never

In today's remote-first world, artifacts are the truth of the company. Every machine-readable trace of work — code, conversations, decisions, documents — is digital exhaust traced back to the person who produced it.

02
Understand
Raw input
#backend "We decided to go with DynamoDB for the session store"
PR #412 "Switched to JWT refresh tokens — see arch review notes"
Standup "I own the deploy runbook for payments, updated it last week"
Slack DM "Latency issue at 50k sessions was the reason"
Extracting
Mapping
Structuring
RK
Ravi K.
Knowledge profile Role: Core Infra Lead
Key Decision
DynamoDB over Postgres
Jan 14 · Latency at 50k sessions · Active
Key Process
Deploy payments service
5 steps · Last confirmed: Feb 19
Relationship
Nika P. (Data)
Frequent overlaps on schema designs
Auth 34 Payments 28 Database 22
Personal Index
When did I update the payments runbook?
You mentioned owning and updating it last week during Standup.

Raw fragments in, structured knowledge out. Brekfuz builds the understanding that used to require three skip-levels and a coffee chat.

03
Traced Answers
@engineering why is the payments canary stalling?
"The canary is blocked on a missing IAM policy update in the deploy config. There's also a recurring 502 on the staging cluster tied to the new retry handler."
RP
Raj P.archived · Slack #payments, Feb 12 4:31 PM
SK
Sarah K. · Jira INFRA-2847
@jules what was your most recent commit to payments and why?
"Last commit was a3f91c2 — patched the Stripe webhook retry logic. The old handler was double-processing events during canary."
JC
Jules C. · GitHub commit a3f91c2, Mar 7

Every claim pins to an exact Slack message, Jira ticket, or commit SHA — down to the timestamp. Knowledge survives offboarding: Raj left six months ago, but his context is still answering questions.

04
Observability
Time
Requester
Query
Lineage
Conf.
Resolution
2h ago
Sarah K.
Why DynamoDB over Postgres?
2 sources
0.87
Confirmed
5h ago
Agent Devin
Why custom retry wrapper in payments?
Slack
0.91
Confirmed
8h ago
Meera R.
Who owns the notification service?
2 sources
0.83
Confirmed
Yesterday
Alex
How to deploy payments service?
2 sources
0.62
Caveated
Your Human API's Response

"Stripe webhook race condition — flag gates the new handler during rollout."

Posted in Slack #payments · Today, 2:14 PM Confirmed
Routed to You

"I found the deploy scripts for the payments service, but the IAM roles config is missing. Should I send what I have to Alex, or do you want to add the IAM details?"

GitHub Commit · Yesterday Caveated

Every query, every source, every resolution — logged with full lineage. Complete visibility into what's being asked, who's answering, and how confident the system is.

05
Two Interfaces
A teammate asks a question
An agent hits undocumented knowledge
# data-requests
TH
Tara H. 2:14 PM
@data why is MRR different in Stripe vs Snowflake?
Data Team API APP

Stripe includes paused subscriptions, while our Snowflake model excludes them as of the Q2 definition update.

JC
via Jules' API · 0.96
2:16 PM
TH
Tara H. 2:16 PM
@aryan can we just use Stripe for the board deck?
AA
Aryan's API2:18 PM
No, Snowflake is the single source of truth for board reporting. Always use Snowflake.
deck-generator → data-team-api
// revenue dashboard building
// agent needs metric definitions
POST /v1/query
{
  "q": "should I use Stripe or Snowflake for MRR",
  "context": "board-deck-generator"
}
// 120ms
{
  "answer": "Use Snowflake. It correctly excludes paused subscriptions per Q2 definition.",
  "confidence": 0.96,
  "sources": 5,
  "employee": null
}
Agent generates correct chart autonomously
Same engine. Two interfaces.

Teammates get attribution. Agents get the answer. The knowledge that was stuck in one person's head now serves both.

Who we build for

Teams where expert knowledge
is of the essence

01
Scaling startups
Teams (often remote!) growing fast enough that coordination breaks before process can catch up.
02
Teams deploying AI agents
Every agent hits the same wall: undocumented decisions in one person's head. We're the context layer they query.
03
Distributed teams
Across time zones where synchronous coordination is no longer viable.
Founders
Sarthak Ahuja
Sarthak Ahuja
Computer Science @ UIUC
Arhan Singhal
Arhan Singhal
Artificial Intelligence @ UPenn
Ethos

The flow of information is throttled by human bandwidth, and we want to change that equation.