Dave superhero mascot Dave superhero mascot The World's First Agentic Staffing Platform1

Give your engineering team
superpowers.

Our vision is to build the world's first Agentic Staffing Platform: hire AI agents by role and seniority, just like staffing. Our first planned agent is SRE Dave, designed for incident response, root cause analysis, and runbook automation with human approval at every step.

We're speaking with a small number of engineering leaders about early design feedback and future Discovery Partner interest.

Apply to Become a Discovery Partner Concept Walkthrough

Built around a Knowledge Platform that learns from your code, docs, logs, and incidents, getting smarter with every resolution

Why Start with SRE?

The industry is obsessed with writing
code faster. The real problem is everything else.

Engineers spend just 52 minutes per day writing code.5 The other 83% goes to maintenance, debugging, incidents, and meetings. 42% of engineering capacity is lost to tech debt alone, costing the industry $85B a year.6 On-call toil is the most painful, measurable problem for engineering teams.

57%

of SREs spend over half their week on toil2

65%

of engineers report burnout from on-call rotations3

$250K

fully-loaded cost of a single SRE per year4

52m
coding / day
83% maintenance, ops & meetings
17% writing code

Today's AI tools optimise for the 17%.
Dave tackles the other 83%.5

Who's on-call at 3 AM when the payment service goes down? Who triages the alert, traces the dependency graph, finds the root cause, and updates the runbook?

That's the work nobody wants to do, and it's where the Dave platform starts. SRE Dave is our first planned agent and the starting point for early design conversations.

The Wedge: SRE Dave

From alert to resolution.
Human always in the loop.

SRE Dave is the first planned agent on the Dave platform. The idea: connect to the observability and developer tools your team already uses, investigate incidents, and learn from every resolution. Human approval required at every step.

1

Ingest

We are designing SRE Dave to work with common observability, source control, and documentation systems to build a Knowledge Platform mapping your services, dependencies, and historical incidents.

Knowledge Platform
2

Investigate

When an alert fires, the intended workflow traces service dependencies, correlates with historical incidents, and identifies the root cause, drawing on everything the Knowledge Platform has learned.

Multi-hop investigation
3

Recommend & Approve

Dave is designed to post a Teams or Slack card with root cause analysis and remediation steps. A human reviews and approves before anything happens.

Human-in-the-loop
4

Execute & Learn

Once approved, the action runs and the outcome feeds back into the Knowledge Platform. The team never solves the same problem twice.

Knowledge Platform
Concept Walkthrough

See the intended workflow

An illustrative incident flow on synthetic data, designed to show the concept: alert fires, Dave investigates, human approves, problem resolved. This animation represents the intended product workflow, not a production deployment.

incidents
The Grand Vision

One platform. One brain.
Every agent gets smarter.

At the heart of Dave is a Knowledge Platform — Dave's brain. The long-term vision: it continuously learns from code, docs, logs, and incidents, building a shared intelligence layer that compounds over time. Future planned agents would inherit everything the platform has already learned.

STARTING POINT
SRE Dave

SRE Dave — First Planned Agent

Incident response, root cause analysis, alert triage, runbook automation. Designed to learn from your code, docs, logs, and every resolution.

Current focus for design exploration
FUTURE ROLE
QA Dave

QA Dave — Second Planned Agent

Test generation, regression detection, environment management. Would leverage the Knowledge Platform's production insights to write tests that actually matter.

FUTURE ROLE
Full Stack Dave

Full Stack Dave — Third Planned Agent

Feature implementation, code review, technical debt reduction. Would be informed by real production behavior and failure patterns the platform brain already knows.

LONG-TERM VISION
UX Dave

UX Dave — Fourth Planned Agent

Design system audits, accessibility checks, user flow analysis. Would use the Knowledge Platform's frontend telemetry and user behavior data to surface UX improvements.

LONG-TERM VISION
PM Dave

PM Dave — Fifth Planned Agent

Product roadmap insights, requirements analysis, stakeholder updates. Would be grounded in the platform's deep understanding of your engineering reality.

& BEYOND
More agents

The long-term vision includes many more agents. Every new Dave would share the same platform brain and make the whole team smarter.

Planned Access Model

How Discovery Partner
access may be scoped.

The following capability concepts illustrate the trust and autonomy model we're designing toward — not a commercial offer, not available for purchase or deployment. They are starting points for design conversations with early partners. Pricing, capacity, and terms are not yet determined.

Read-Only Investigation
SRE Dave
Coming Soon

Read-only observation and documentation. The intended entry point: safe, observable, and entirely human-approved.

  • Read-only access to logs and metrics
  • Root cause analysis and recommendations
  • Drafts PRs for runbook and doc updates
  • 100% human approval required
  • Incident investigation capacity (TBD)
  • Teams/Slack Adaptive Card notifications
Can Dave change code?No
Can Dave change infra?No
Human approval required?Always
Express Interest
Draft Remediation
SRE Dave
Coming Soon

Deeper correlation, code-level PR drafts, and higher investigation capacity. Everything in Read-Only Investigation, plus:

  • Multi-hop dependency correlation
  • Drafts code and infrastructure PRs
  • Deployment correlation analysis
  • Higher incident investigation capacity (TBD)
  • Planned: hands-on design partner support
Can Dave change code?Drafts only
Can Dave change infra?No
Human approval required?Always
Express Interest
Approved Safe Actions
SRE Dave
Coming Soon

Pre-approved safe pattern execution and maximum autonomy. Everything in Draft Remediation, plus:

  • Executes pre-approved safe patterns (with approval gate)
  • Unlimited incident investigation volume (TBD)
  • Meeting Intelligence module (planned)
  • Custom Knowledge Platform tuning
  • Planned: dedicated design partner support
Can Dave change code?Approved patterns
Can Dave change infra?Approved patterns
Human approval required?Always
Express Interest
Future Module — Planned Extension
Meeting Intelligence
Planned Module Envisioned for Approved Safe Actions scope
Intended to work with recording-capable meeting platforms
Google Meet Zoom Microsoft Teams
  • Auto-schedules PIR meetings: creates and schedules post-incident reviews after P1/P2 resolution, no manual coordination needed.
  • Silent observer during incidents: joins incident calls and absorbs context in real time, zero interference while engineers focus on mitigation.
  • Recording analysis and action items: processes meeting recordings, extracts decisions and RCA insights, and feeds learnings back into the Knowledge Platform.
  • PIR meeting participation: prompts RCA questions, flags patterns from historical data, and suggests preventive actions.
  • Auto-generated summaries: pushes meeting summaries with action items to Slack and Jira, linked to related incidents.
Design principle: Dave stays silent during live incident calls. Intelligence surfaces during PIRs and asynchronously via Slack/Jira, where it adds the most value without getting in the way.

These tiers represent planned product design concepts, not commercial offers. Pricing, capacity limits, and availability are not yet determined. A fully-loaded SRE costs $150K–$250K/year — the long-term vision targets a fraction of that cost.

Trust & Security

Designed for the "AI blast radius"
conversation with your CISO.

Dave proposes, humans decide. Every action is scoped, logged, and reversible.

Read-Only-First Design

The intended early-access model begins with read-only scope. Dave observes, analyzes, and recommends. Nothing gets touched. Expanded scope is a deliberate, future decision, not a default.

Least-Privilege Access

Strict path allowlists. Dave only accesses the repositories, services, and environments you explicitly authorize. No lateral movement, no scope creep.

Mandatory Approval Gates

Every action above read-only requires explicit human approval via Teams or Slack Adaptive Cards. No destructive actions without human sign-off. Ever.

No Data Exfiltration

Your code and logs stay within your perimeter. Dave's RAG pipeline processes data in your environment. Nothing is sent to third-party training sets.

Full Audit Trail

Every investigation, recommendation, and action is logged with timestamps, reasoning chains, and approval records. Complete traceability for compliance reviews.

Building Toward Compliance

We are designing toward formal compliance requirements. Planned controls include mandatory approval gates, least-privilege access, encryption at rest and in transit, role-based access controls, and a self-hosted deployment option.

Built by engineers who've been on-call at 3 AM.

Read our story

Frequently asked questions

The staffing industry solved scaling for humans decades ago: you define the role, the scope, and the trust level, then you hire. Dave is a concept-stage platform that applies the same model to AI agents. Instead of buying a generic "AI DevOps tool," the vision is to explore role-based agent concepts for engineering teams — scoped by capability and trust, at a fraction of the cost of human headcount. Every agent on the platform is designed to share the same Knowledge Platform (Dave's brain), which is intended to get smarter with every interaction. SRE Dave is the first planned agent, with more roles in the long-term vision.
Dave is built on a purpose-built Knowledge Platform that combines semantic search with service dependency mapping to eliminate hallucinations and ground every recommendation in your actual systems. The platform uses a multi-step investigation pipeline with tiered AI reasoning: fast classification for triage, expert-level analysis for root cause. We didn't build a custom LLM from scratch; we built a purpose-built orchestration and knowledge layer on top of the best foundation models available.
Not without your explicit approval. In the Read-Only Investigation concept, Dave is entirely read-only. In the Draft Remediation concept, Dave would draft code and infrastructure PRs that require human review. Even in the Approved Safe Actions concept, Dave would only execute pre-approved safe patterns that your team has explicitly allowlisted. Every action above read-only is designed to go through a mandatory Teams/Slack approval gate. Dave proposes, humans decide. These are planned design constraints, not currently deployed behaviors.
Two reasons. First, on-call toil is the most painful, measurable problem for engineering teams: 57% of SREs spend over half their week on it, and 65% report burnout. The ROI is immediate and quantifiable. Second, starting from operations means the Knowledge Platform learns how your production systems actually work: real dependencies, real failure modes, real runbooks. That context compounds and makes every future Dave role (QA, Dev, PM) dramatically more effective.
Our intended early-access model begins with read-only scope: Dave would connect to monitoring, source control, and documentation tools via API, webhook, or OAuth, build the Knowledge Platform, and investigate incidents alongside your team. It is being designed to work with any cloud provider and any observability stack. Future design partners may receive hands-on support as the program evolves. We are currently speaking with engineering leaders who want to help shape the product through direct feedback, not purchase a live product today.
Most AI developer tools focus on writing new code faster, targeting the left side of the software development lifecycle. But engineers only spend 17% of their time writing code. Dave starts from the right side: operations, where 42% of engineering time is lost to maintenance and tech debt. We're not a code generation tool; we're a staffing platform that starts where the toil is highest, with a knowledge platform purpose-built for multi-hop incident investigation across your entire stack.
Dave is being designed to work with tools that support API, webhook, or OAuth, including categories such as observability (e.g., Datadog, Azure App Insights, PagerDuty, Grafana), source control (e.g., GitHub, GitLab, Bitbucket), documentation (e.g., Confluence, Notion), and notification channels for approval gates (e.g., Microsoft Teams / Slack). The platform is intended to be cloud-agnostic across AWS, Azure, GCP, or any combination.
No. The intended early-access model begins in read-only mode so teams can evaluate Dave's recommendations, workflows, and outputs safely. Dave is designed to observe, investigate, and document. It would not modify code, infrastructure, or configurations. Any expanded scope would be a deliberate decision, not a default.
Never without approval. In the Approved Safe Actions design concept, infrastructure changes would be restricted to pre-approved safe patterns explicitly allowlisted by the team, and every action would still go through an approval gate via Teams or Slack. There is no path for Dave to make autonomous, unapproved changes. The approval gate is a hard design constraint, not a configurable option.
Think of it as a planned progression of trust and capability. The Read-Only Investigation concept is designed to observe and document: read-only access to logs and metrics, with root cause analysis and recommendations. Draft Remediation is intended to add the ability to draft code and infrastructure PRs for human review. Approved Safe Actions is designed to execute pre-approved safe workflows, but would still require human approval for anything outside those allowlisted patterns. These are planned design concepts, not currently available products.

Apply to become a Discovery Partner.

We're speaking with a small number of engineering leaders who want to help shape SRE Dave's workflows, trust boundaries, and approval model. Discovery Partners provide structured feedback — this is not an offer of software deployment or a commercial pilot.

✓ Concept-stage — no deployment ✓ Feedback calls & workflow review ✓ Recorded concept walkthrough available on request

If there's fit, we may reach out to explore a Discovery Partner conversation — no commitment required

We respect your privacy. Your email will never be shared with third parties or used for spam.