Cortex

sitemate-product-vision

8 min read

Sitemate Product Vision

Summary

The complete product strategy arc from Hartley’s blog series (Oct 2025 – Jan 2026). Starts with the mission (why Sitemate exists), defines where to compete (front line, not back office), how to build (vertical software with no-code principles), and what the AI-era product looks like (zero-click forms). Each layer builds on the one below it.

Key Points

Mission: Multiply the Engineering Power of the Built World

  • Not “take people digital” — unshackle the engineering workforce from burnout and administrative burden
  • The built world is at a multi-pronged crisis: $15T infrastructure gap, rising costs, 98% of large projects over budget/late, crumbling existing infrastructure, rising regulation
  • All pressure vectors converge on the people doing the work — “the ones who work in the dark, work in the cold, and work in the silence”
  • Origin: Hartley spent 30%+ of his time as a civil engineer on mindless paperwork. “This was the moment that Sitemate began to pull itself out of me.”

Where to compete: Take AI to the Front Line

  • Three-layer map of where AI applies:
    • Back Office (APIs, MCP, data warehousing) — red ocean, horizontal providers will outspend everyone. Sitemate stance: integrate cleanly, don’t compete here.
    • Messy Middle (safety/quality/maintenance workflows) — horizontal vs vertical battleground. Accuracy and traceability matter more than cleverness. Hardest layer to win.
    • Front Line (mobile, cross-party, hostile environments) — where truth is created. Sitemate’s differentiator.
  • “AI becomes defensible when it helps create high-integrity truth in an otherwise messy reality, at volume, with near-zero friction.”
  • “If we win the truth layer, we become the system other systems rely on.”
  • Staging order: front line first → back office next → messy middle after

How to build: Vertical Software with No-Code Principles

  • “Opinionated, yet configurable” — opinionated about core built world problems, flexible about the last mile
  • Answer to “build vs buy?” is “do both — buy, then build”
  • Lego block architecture: each field, rule, workflow step is a composable primitive. Adding one block increases possibility space non-linearly.
  • No-code is “grease in the gears” across three dimensions:
    • Use case accessibility — last mile varies by client, contract, region
    • Segment accessibility — VSB (speed) → mid-market (flexibility) → enterprise (governance). “Entry level simplicity, enterprise power behind gradual reveal.”
    • Geographic accessibility — processes rhyme but don’t repeat across regions
  • Primary benefit: friction removal. Second-order: mounting loss (switching costs from accumulated configuration)
  • Vibe coding is accelerant, not threat — gets 80-90%, last 10-20% (permissions, governance, auditability, offline, scale) is where Sitemate wins
  • HP ↔ Will Smith spectrum: ultra-flex vs opinionated workflows. Goldilocks zone in the middle.

The AI product: Zero Click Forms

  • “What if forms didn’t need to be filled? What if they could be produced as a side effect of work happening?”
  • The built world “double jump” — skip digital form filling, go straight from paper to AI-generated records (like African markets skipping PCs for smartphones)
  • Multi-modal inputs: voice + photos + video + context → structured auditable records
  • Target flows:
    1. Daily Reporting — voice conversation in the truck → completed report
    2. Inspections — voice monologue + photos while walking → completed form + NCRs
    3. Time Capture — passive signals + lightweight confirmation → timesheets
    4. Inductions — conversational agent in any language → completed record (solves for illiterate workforce)
    5. Morning Briefings — supervisor hits record → brief, attendance, hazards, actions
  • Prerequisites: context-aware agent, good questions (not generic), auditable outputs, correction loop, fits real environment (gloves, noise, sun glare)
  • Two simultaneous goals: drive usage up (more use cases) + drive time-to-complete down (fewer clicks)

Four competitive buckets

  1. Status quo (paperwork, word, excel, PDFs) — still the vast majority
  2. Niche apps (safety, timesheets, quality)
  3. All-in-ones (everything in one UI)
  4. In-house builds (internal devs or agencies)
  • Sitemate is a hybrid of all four

Platform evolution roadmap (from Q2 Leadership Kickoff)

  • Three clear capability jumps:
    1. V1: Zero-click forms — AI templates + form fill within Sitemate products (current)
    2. V2: Conversational data capture — agent extends outside products
    3. V3: Visual data capture integration — mid-market/enterprise focus
  • Core equation: more data → better forms → more usage → more data (flywheel)
  • Agent will enable proactive usage and potentially eliminate much of CS onboarding

External validation: API-first / headless-first mandate (Aaron Levie, Box, Apr 2026)

  • Button-heavy apps lose value in an agent world; API-first platforms win
  • APIs must carry embedded business logic (not raw DB access) for agents to operate safely — mirrors Sitemate’s “opinionated, yet configurable” stance at the API layer
  • Box now does more API calls than UI calls — agents are a force multiplier, not a threat, if the product is designed headless-first
  • “The workflow needs to be redesigned for agents, not for people” — front-line zero-click forms are a workflow-for-agents play, not a UI modernization play
  • Enterprise AI adoption will lag SV expectations by years (fragmented systems, liability, governance) — reinforces Sitemate’s staging: win the truth layer at the front line first, the back-office agent market will come to us

Three-layer stack & sequencing (refined at 2026-05-05 Managers Allhands)

  • Refined three-layer model:
    • Frontline (bottom): fan-out from one capture mode (forms) to many — photos, videos, voice clips, phone calls, emails, SMS, notes. Most AI Platform investment goes here through June 1st launch.
    • Middle (“digital employee”): workflows where most usage in dash pivot isn’t done by a human — 24/7 safety manager, fleet manager, etc. This is the V2.0 direction.
    • Back office (top): MCP (3 beta connectors live, productionising), Open API, exports — defensive against rising customer demand for their own data for custom AI use. API usage already skyrocketing.
  • Sequencing: fan out frontline UX → open up back office (low friction, rate-limited) → then make the call on which digital employee to build first.
  • Mobile ring architecture (new framing for Sean’s question): old world = nested rings from system admin (core) → web/mobile staff → workforce app. New shift adds an outer ring: don’t need to be in the product at all — phone call, voice, SMS, native language. Sitemate mobile web sits as a bridge between phone and full app — three rungs at the workforce level.

Agent 1.0 vs 2.0

  • 1.0 (June launch): form fill + AI templates. Intentionally generic across the platform — improves the two most common UX journeys, given to all users, no commercial change. Goal: broad foundational awareness.
  • 2.0 (July–Sep build, October launch): functional digital employees (e.g. agent fleet, agent safety) that operate inside the SaaS rails for the customer. Larger commercial changes likely — offering shifts from “software a human operates” to “digital employee that runs on the rails, collaborates side by side.”
  • Single-function focus from Q3: ~70% of PDE resources behind one function (likely safety based on data Peggy is collating). Build the functional agent + the SaaS infrastructure it needs, in parallel. 3–6–9 months per function before moving to the next.

Commercial model evolution

  • Sitemate Start: new billable item launching ~first week of June. For customers with multiple workspaces needing shared data management logic. Volume of ops will be lower than Dashpivot but may extend into SMB based on Peggy’s research.
  • No commercial change for 1.0 launch — intentional, to make zero-click the default, not a paid add-on.
  • 2.0 may introduce larger commercial changes once early-launch hypotheses are verified; June all-hands will lock in team shuffle/reorg.

AI safety controls philosophy (vs certifications)

  • “If we were a customer, what controls would we want?” — three levels of toggle going live with launch:
    1. Workspace-level on/off
    2. Template-level on/off (e.g. on for daily reports, off for safety checklists)
    3. Per-checklist-item human-required toggle
  • Hypothesis: these three levels deflect ~90% of objections in the first month.
  • Model selection = next priority. Certification (ISO AI / similar) = third priority — concern that AI standards evolve faster than certifications can keep up.

Safety-first pivot (post-June 2026)

  • Multi-category approach creating dilution across PDE and GTM
  • Safety identified as clear “king category” across all data points
  • Post-June 1st: spear architecture — lead with safety, other categories follow
  • 80% of PDE resources reallocating to safety
  • GTM implication: AEs lead with safety, AMs handle cross-sell expansion
  • Q3+ likely company-wide alignment around single category

Sources

Last Updated

2026-05-06

Graph View

Backlinks